Power Allocation Games in Wireless Networks of Multi-antenna Terminals

Size: px
Start display at page:

Download "Power Allocation Games in Wireless Networks of Multi-antenna Terminals"

Transcription

1 Power Aocation Games in Wireess Networs of Muti-antenna Terminas Eena-Veronica Bemega, Samson Lasauce, Mérouane Debbah, Marc Jungers, and Juien Dumont arxiv:0.4598v2 [cs.it] 23 Nov 200 Abstract We consider wireess networs that can be modeed by mutipe access channes in which a the terminas are equipped with mutipe antennas. The propagation mode used to account for the effects of transmit and receive antenna correations is the unitary-invariant-unitary mode, which is one of the most genera modes avaiabe in the iterature. In this context, we introduce and anayze two resource aocation games. In both games, the mobie stations sefishy choose their power aocation poicies in order to maximize their individua upin transmission rates; in particuar they can ignore some specified centraized poicies. In the first game considered, the base station impements successive interference canceation (SIC) and each mobie station chooses his best space-time power aocation scheme; here, a coordination mechanism is used to indicate to the users the order in which the receiver appies SIC. In the second framewor, the base station is assumed to impement singe-user decoding. For these two games a thorough anaysis of the Nash equiibrium is provided: the existence and uniqueness issues are addressed; the corresponding power aocation poicies are determined by expoiting random matrix theory; the sum-rate efficiency of the equiibrium is studied anayticay in the ow and high signa-to-noise ratio regimes and by simuations in more typica scenarios. Simuations show that, in particuar, the sum-rate efficiency is high for the type of systems investigated and the performance oss due to the use of the proposed suboptimum coordination mechanism is very sma. Index Terms theory. MIMO, MAC, non-cooperative games, Nash equiibrium, power aocation, price of anarchy, random matrix E. V. Bemega and S. Lasauce are with LSS (joint ab of CNRS, Supéec, Univ. Paris-Sud ), Supéec, Pateau du Mouon, 992 Gifsur-Yvette Cedex, France, {bemega,asauce}@ss.supeec.fr; M. Debbah is with the Acate-Lucent Chair on Fexibe Radio, Supéec, Pateau du Mouon, 992 Gif-sur-Yvette Cedex, France, merouane.debbah@supeec.fr; M. Jungers is with CRAN, Nancy-Université, CNRS, 2 avenue de a Forêt de Haye, 5456 Vandoeuvre-es-Nancy, France, marc.jungers@cran.uhp-nancy.fr; J. Dumont is with Lycée Chapta, 45 Bouevard des Batignoes, Paris, France, dumont@crans.org

2 2 I. INTRODUCTION In this paper, we consider the upin of a decentraized networ of severa mobie stations (MS) and one base station (BS). This type of networ is commony referred to as the decentraized mutipe access channe (MAC). The networ is said to be decentraized in the sense that each user can freey choose his power aocation (PA) poicy in order to sefishy maximize a certain individua performance criterion, which is caed utiity or payoff. This means that, even if the the BS broadcasts some specified poicies, every user is free to ignore the poicy intended for him if the atter does not maximize his performance criterion. To the best of the authors nowedge, the probem of decentraized PA in wireess networs has been propery formaized for the first time in [], [2]. Interestingy, this probem can be formuated quite naturay as a non-cooperative game with different performance criteria (utiities) such as the carrier-to-interference ratio [3], aggregate throughput [4] or energy efficiency [5], [6]. In this paper, we assume that the users want to maximize information-theoretic utiities and more precisey their Shannon transmission rates. Indeed, the point of view adopted here is cose to the one proposed by the authors of [7] for DSL (digita subscriber ines) systems, which are modeed as a parae interference channe; [8] for the singe input singe output (SISO) and singe input mutipe output (SIMO) fast fading MACs with goba CSIR and goba CSIT (Channe State Information at the Receiver/Transmitters); [9] for MIMO (Mutipe Input Mutipe Output) MACs with goba CSIR, channe distribution information at the transmitters (goba CDIT) and singe-user decoding (SUD) at the receivers; [0], [] for Gaussian MIMO interference channes with goba CSIR and oca CSIT and, by definition of the conventiona interference channe [2], SUD at the receivers. Note that reference [3] where the authors considered Gaussian MIMO MACs with neither CSIT nor CDIT differs from our approach and that of [7], [8], [9], [0], [] because in [3] the MIMO MAC is seen as a two-payer zero-sum game where the first payer is the group of transmitters and the second payer is the set of MIMO sub-channes. The cosest wors to the wor presented here are [9] and [4]. Athough this paper is in part based on these wors, it sti provides significant contributions w.r.t. to them, as expained beow. In [9], the authors consider MIMO mutipe access channes and assume SUD at the BS; the authors formuate the PA probem into a team game in which each user chooses his PA to maximize the networ sum-rate. In [4], the same type of decentraized networs is considered but SIC is assumed at the BS. As each user needs to now his decoding ran in order to adapt his PA poicy to maximize his individua transmission rate, a coordination mechanism has to be introduced: the coordination signa precisey indicates to a the users the decoding order used by the receiver. The present paper differs from these two contributions on at east four

3 3 important technica points: (i) when SUD is assumed, the PA game is not formuated as a team game but as a non-cooperative one; (ii) we expoit severa proof techniques that are different from [9]; (iii) whie [9] and [4] assume a Kronecer propagation mode with common receive correation we assume here a more genera mode, the unitary-invariant-unitary (UIU) propagation mode introduced by [2], for which the users can have different receive antenna correation profies. This is usefu in practice since, for instance, it aows one to study propagation scenarios where some users can be in ine of sight with the BS (the receive antenna are strongy correated) whereas other users can be surrounded by many obstaces, which can strongy decorreate the receive antennas for these users; (iv) whie the authors of [4] restricted their attention to either a purey spatia PA probem or a purey tempora PA probem, we tace here the genera space-time PA probem. In this context, our main objective is to study the equiibrium of two power aocation games associated with the two types of decoding schemes aforementioned (namey SIC and SUD). The motivation for this is that the existence of an equiibrium aows networ designers to predict, with a certain degree of stabiity, the effective operating state(s) of the networ. Ceary, in our context, uniqueness is a desirabe feature of the equiibrium. As it wi be seen, it is possibe to prove the existence in both games under investigation. Uniqueness is proven in the case of SUD whie it is conjectured for the case of SIC. In order to estabish the corresponding resuts, the paper is structured as foows. After presenting the genera system mode in Sec. II, we anayze in detai the space-time PA game when SIC and a corresponding coordination mechanism are assumed (Sec. III). For this game, the existence and uniqueness of the NE are proven and the equiibrium is determined by expoiting random matrix theory when the numbers of antennas are sufficienty arge. Its sum-rate efficiency is aso anayzed. In Sec. IV, we anayze the case of SUD since this decoding scheme, athough suboptima in terms of performance (even in the case of a networ with singe-antenna terminas), has some features that can be found desirabe in some contexts: the receiver compexity is ow, there is no need for a coordination signa, there is no propagation error since the data fows are decoded in parae and not successivey and aso it is intrinsicay fair. To anayze the case of the SUD-based PA game, we wi foow the same steps as in Sec. III and we wi see that, the equiibrium anaysis can be deduced, to a arge extent, from the SIC case. Numerica resuts are provided in Sec. V to iustrate our theoretica anaysis and to better assess the sum-rate efficiency of the considered games. Sec. VI corresponds to the concusion. II. SYSTEM MODEL We assume a MAC with arbitrary number of users, K 2. Regarding the origina definition of the MAC by [5] and [6], the system under consideration has two common features: a transmitters send at once and at

4 4 different rates over the entire bandwidth, and the transmitters are using good codes in the sense of the Shannon rate. Our system differs from [5][6] in the sense that mutipe antennas are considered at the termina nodes, channes vary over time and the BS does not dictate the PA poicies to the MSs. Aso, we assume the existence of coordination signa which is perfecty nown to a the terminas. If the coordination signa is generated by the BS itsef, this induces a certain cost in terms of downin signaing but the distribution of the coordination signa can then be optimized. On the other hand, if the coordination signa comes from an externa source, e.g., an FM transmitter, the MSs can acquire their coordination signa for free in terms of downin signaing. However this generay invoves a certain sub-optimaity in terms of upin rate. In both cases, the coordination signa wi be represented by a random variabe denoted by S S. Since we study the K user MAC, S = {0,,...,K!} is a K!+-eement aphabet. When the reaization is in {,...,K!}, the BS appies SIC with a certain decoding order (game ). When S = 0 the BS aways appies SUD (game 2), where a users are decoded simutaneousy (no interference canceation). In a rea wireess system the frequency at which the reaizations woud be drawn woud be roughy proportiona to the reciproca of the channe coherence time (i.e., /T coh ). Note that the proposed coordination mechanism is suboptima because it does not depend on the reaizations of the channe matrices. We wi see that the corresponding performance oss is in fact very sma. We wi further consider that each mobie station is equipped with n t antennas whereas the base station has n r antennas (thus we assume the same number of transmitting antennas for a the users). In our anaysis, the fat fading channe matrices of the different ins vary from symbo vector (or space-time codeword) to symbo vector. We assume that the receiver nows a the channe matrices (CSIR) whereas each transmitter has ony access to the statistics of the different channes (CDIT). The equivaent baseband signa received by the base station can be written as: Y (s) (τ) = K = H (τ)x (s) (τ)+z(s) (τ), () wherex (s) (τ) is the n t-dimensiona coumn vector of symbos transmitted by user at time τ for the reaization s S of the coordination signa, H (τ) C nr nt is the channe matrix (stationary and ergodic process) of user and Z (s) (τ) is a n r -dimensiona compex white Gaussian noise distributed as N(0,σ 2 I nr ). For the sae of carity we wi omit the time index τ from our notations. In order to tae into account the antenna correation effects at the transmitters and receiver, we wi assume the different channe matrices to be structured according to the unitary-independent-unitary mode introduced in [2]: {,...,K}, H = V H W, (2)

5 5 where V and W are deterministic unitary matrices that aow one to tae into consideration the correation effects at the receiver and transmitter. Aso H is an n r n t matrix whose entries are zero-mean independent compex Gaussian random variabes with an arbitrary profie of variances, such that E H (i,j) 2 = σ(i,j) n t. The Kronecer propagation mode for which the channe transfer matrices factorizes as H = R /2 specia case of the UIU mode where the profie of variances is separabe i.e., E H (i,j) 2 = d(r) Θ T /2 (i)d(t)(j) n t is a, with for each : Θ is a random matrix with zero-mean i.i.d. entries, T is the transmit antenna correation matrix, R is the receive antenna correation matrix, {d (T) (j)} j {,...,nt} and {d (R) (i)} i {,...,nr} are their associated eigenvaues. In this paper we wi consider that V = V for a users. The reason for assuming this wi be made cearer a itte further. In spite of this simpification, we wi sti be abe to dea with some usefu scenarios where the users see different propagation conditions in terms of receive antenna correation. III. SUCCESSIVE INTERFERENCE CANCELLATION When SIC is assumed at the BS, the strategy of user {,2,...,K}, consists in choosing the best vector ( ) [ ] of precoding matrices Q = Q (),Q(2),...,Q(K!) where Q (s) = E X (s) X(s),H, for s S, in the sense of his utiity function. For carity sae, we wi introduce another notation which wi be used in the remaining of this section to repace the reaization s of the coordination signa. We denote by P K the set of a possibe permutations of K eements, such that π P denotes a certain decoding order for the K users and π() denotes the ran of user K and π P K denotes the inverse permutation (i.e. π (π()) = ) such that π (r) denotes the index of the user that is decoded with ran r K. We denote by p π [0,] the probabiity that the receiver impements the decoding order π P K, which means that p π =. At ast note that there is a one-to-one mapping between the set of reaizations of the coordination signa S and the set of permutations P K, i.e. ξ : S P such that ξ( ) is a bijective function. This is the reason why the index s can be repaced with the index π without introducing any ambiguity or oss of generaity. The vector of precoding matrices can ( ) be denoted by Q = Q (π) and the utiity function can be written as: where R (π) (Q(π),Q(π) u SIC (Q,Q ) = p π R (π) (Q(π),Q(π) ) (3) ) = Eog 2 I+ρH Q (π) HH +ρ K (π) H Q (π) H H Eog 2 I+ρ K (π) H Q (π) H H (4)

6 6 with ρ = σ 2 and K (π) = { K π() π()} represents, for a given decoding order π, the subset of users that wi be decoded after user. Aso, we use the standard notation, which stands for the other payers than. An important point to mention here is the power constraint under which the utiities are maximized. Indeed for user {,...,K}, the strategy set is defined as foows: { } ( ) A SIC = Q = Q (π) π P K,Q (π) 0, p π Tr(Q (π) ) n tp. (5) In order to tace the existence and uniqueness issues for Nash equiibria in the genera space-time PA game, we expoit and extend the resuts from Rosen [7], which we wi briefy state here beow in order to mae this paper sufficienty sef-contained. Theorem : [7] Let G = (K,{A } K,{u } K ) be a game where K = {,...,K} is the set of payers, A,...,A K the corresponding sets of strategies andu,...,u the utiities of the different payers. If the foowing three conditions are satisfied: (i) each u is continuous in the a the strategies a j A j, j K; (ii) each u is concave in a A ; (iii) A,...,A K are compact and convex sets; then G has at east one NE. Theorem 2: [7] Consider the K-payer concave game of Theorem. If the foowing (diagonay strict concavity) condition is met: for a K and for a (a,a ) A2 such that there exists at east one index K j K for which a j a j, (a [ a )T a u (a,a ) a u (a,a )] > 0; then the uniqueness of the NE is insured. = In the space-time power aocation game under investigation, the obtained resuts are stated in the foowing theorem. Theorem 3: [Existence of an NE] The joint space-time power aocation game described by: the set of payers K; the sets of actions A SIC and the utiity functions u SIC (Q,Q ) given in (3), has a Nash equiibrium. Proof: It is quite easy to prove that the strategy sets A SIC are convex and compact sets and that the utiity functions u SIC (Q,Q ) are concave w.r.t. Q and continuous w.r.t. to (Q,Q ) and by Theorem at east one Nash equiibrium exists. For more detais, the reader is referred to Appendix A. Theorem 4: [Sufficient condition for uniqueness] If the foowing condition is met K ( Tr {(Q (π) Q (π) ) u SIC (Q,Q ) Q (π) = ) ) Q (π) u SIC (Q,Q ) )} > 0 (6) ( ( for a Q = Q (π),q = Q (π) A SIC such that (Q π P,...,Q K ) (Q,...,Q K ), then the K Nash equiibrium in the power aocation game of Theorem 3 is unique. This theorem corresponds to the matrix generaization of the diagonay strict concavity (DSC) condition of [7] and is proven in Appendix B. To now whether this condition is verified or not in the MIMO MAC one

7 needs to re-write it in a more expoitabe manner. It can be checed that C expresses as C = for each π P K, T π is given by: where A (π) r T π = K Tr = K = E Tr r= ( ( = E {(Q (π) )[ Q (π) Q (π) R (π) (Q(π),Q (π) {ρh π (r)(q (π) π (r) Q(π) π (r) )HH π (r) I+ρH π (r)q (π) π (r) HH π (r) +ρ K I+ρH π (r)q (π) π (r) HH π (r) +ρ K K Tr r= E[F π (H)] ( A (π) r ) ( A (π) r I+ = ρh π (r)q (π) π (r) HH π (r), A(π) r ) Q (π) R (π) (Q(π) H π (s)q (π) π (s) HH π (s)) s=r+ H π (s)q (π) π (s) HH π (s) s=r+ K s=r A (π) s ) ( I+ K s=r = ρh π (r)q (π) π (r) HH π (r) A (π) s ) ) 7 p π T π where,q (π) ) ]} (7) and the users have been ordered using their decoding ran rather than their index. Notice that since the expectation operator is inear we can switch between the trace and expectation. ) Let us denote by H = [H,...,H K ], Q = (Q ) K,, Q = (Q ) K, = (Q (π) Q(π) ) ) ),, K Q(π), K A = (A (π), π P A = (A (π). K, K, K In order to prove that the DSC condition hods we have to prove that for a Q Q we have C > 0. (Q (π) Let us give a very usefu resut. Lemma : For any positive definite matrices A, B, and any positive semi-definite matrices A i, B i, i {2,...,K}, we have that K i Tr (A i B i ) i= j= B j i where the equaity hods if and ony if A j = B j for a j {,...,K} j= A j 0 (8) The proof can be found in [27], for K = 2, and in [28] for arbitrary K 2. Using this resut, we can prove that for any channe reaization, any Q,Q and any π P K : ( F π (H) = Tr A (π) r ) ( A (π) r I+ K s=r A (π) s ) ( I+ K s=r A (π) s ) = 0 (9)

8 8 impying that T π 0 and that C 0. Let us consider now two arbitrary covariance matrices such that Q Q. This means that there is at east one decoding order ϑ P K such that Q (ϑ) Q (ϑ). We wi prove that T ϑ > 0 which wi impy the desired resut C > 0. Remar: Assuming that ran(h H H ) = n t, for a K, and n t n r +, then Q Q impies that A A. This means that for any channe reaization we have F ϑ (H) > 0 which impies directy T ϑ > 0 and C > 0. For the genera proof, et us define the foowing sets: } A H (Q (ϑ),q (ϑ) (ϑ) ) = {H D H K : H (Q (ϑ) Q )H H = } 0 (0) à H (Q (ϑ),q (ϑ) ) = {H D H K : H (Q (ϑ) Q (ϑ) )H H 0 We now that: T ϑ = E[F ϑ (H)] = F ϑ (H)L(H)dH D H = F ϑ (H)L(H)dH+ A H(Q (ϑ),q (ϑ) ) = F ϑ (H)L(H)dH à H(Q (ϑ),q (ϑ) ) à H(Q (ϑ),q (ϑ) ) F ϑ (H)L(H)dH where L(H) > 0 stands for the p.d.f. of H D H C nr Knt. The second equaity foows since D H = A H (Q (ϑ),q (ϑ) ) ÃH(Q (ϑ),q (ϑ) ). The third equaity foows since for a H A H (Q (ϑ),q (ϑ) ) we have that F ϑ (H) = 0 from Lemma 8. We now that for a H ÃH(Q (ϑ),q (ϑ) ) we have that F ϑ (H) > 0. It suffices to prove that à H (Q (ϑ),q (ϑ) ) is a subset of non-zero Lebesgue measure to impy that T ϑ > 0 and thus that C > 0. It turns out that we can prove the existence of a compact set U H ÃH(Q (ϑ),q (ϑ) ) for arbitrary Q (ϑ) Q (ϑ). Thus, we have the desired resut C > 0. Determination of the Nash equiibrium. In order to find the optima covariance matrices, we proceed in the same way as described in [9]. First we wi focus on the optima eigenvectors and then we wi determine the optima eigenvaues by approximating the utiity functions under the arge system assumption. Theorem 5: [Optima eigenvectors] For a K, Q A SIC there is no oss of optimaity by imposing the structure Q = (Q (π) ) π PK, Q (π) = W P (π) W H, in the sense that: where S SIC = Diag(P (π) (),...,P (π) (n t )). max Q A SIC u SIC (Q,Q ) = max Q S SIC { Q = (Q (π) ) π P A SIC Q (π) = W P (π) WH u SIC (Q,Q ), () }, s S, mode from (2) and P (s) =

9 9 The detaied proof of this resut is given in Appendix C. This resut, athough easy to obtain, it is instrumenta in our context for two reasons. First, the search of the optimum precoding matrices bois down to the search of the eigenvaues of these matrices. Second, as the optimum eigenvectors are nown, avaiabe resuts in random matrix theory can be expoited to find an accurate approximation of these eigenvaues. Indeed, the eigenvaues are not easy to find in the finite setting. They might be found using numerica techniques based on extensive search. Here, our approach consists in approximating the utiities in order to obtain expressions which are not ony easier to interpret but aso easier to be optimized w.r.t. the eigenvaues of the precoding matrices. The ey idea is to approximate the different transmission rates by their arge-system equivaent in the regime of arge number of antennas. The corresponding approximates can be found to be accurate even for reativey sma number of antennas (see e.g., [8][9] for more detais). Since we have assumed V = V, we can expoit the resuts in [20][2] for singe-user MIMO channes, assuming the asymptotic regime in terms of the number of antennas: n r, n t, corresponding approximated utiity for user is: n r n t β. The where R (π) (P(π),P(π) ) = n r ũ SIC ({P (π) } K, ) = n r n r K (π) n r i= n t {} j= og 2 + K (π) n r K (π) n r n t {} j= n t j= og 2 + n r i= n r K (π) n r j= p R(π) π (P(π),P(π) ( ) og 2 +(N (π) +)ρp (π) (j)γ (π) (j) + (N (π) +)n t K (π) γ (π) (j)δ (π) (j)og 2 e {} j= ) (2) n t og 2 ( +N (π) ρp (π) (j)φ (π) (j) N (π) n t K (π) φ (π) (j)ψ (π) (j)og 2 e n t j= σ (i,j)δ (π) (j) ) σ (i,j)ψ (π) (j) + (3)

10 0 where N (π) of: = K (π) and the parameters γ(π) (j) and δ(π) (j) j {,...,n t}, K, π P K are the soutions γ (π) (j) = δ (π) (j) = j {,...,n t }, K (π) {} : n r (N (π) +)n t i= + (N (π) +)nt (N (π) +)ρp (π) (j) +(N (π) +)ρp (π) (j)γ (π) (j), σ (i,j) n t σ r (i,m)δ r (π) (m) r K (π) {} m= and φ (π) (j), ψ (π) (j), j {,...,n t } and π P K are the unique soutions of the foowing system: j {,...,n t }, K (π) φ (π) (j) = ψ (π) (j) = N (π) n t The corresponding water-fiing soution is: n r i= : + N (π) nt N (π) ρp π) (j) +N (π) ρp (π) (j)φ (π) P (π),ne (j) = [ n2n r λ r K (π) (j). σ (i,j) n t m= σ r (i,m)ψ (π) r (m) N (π) ργ (π) (j) where λ 0 is the Lagrangian mutipier tuned in order to meet the power constraint: n t j= [ p π n2n r λ N (π) ργ (π) (j) (4) (5) ] +, (6) ] + = n t P. Note that to sove the system of equations given above, we can use the same iterative power aocation agorithm as the one described in [9]. At this point, an important point has to be mentioned. The existence and uniqueness issues have be anayzed in the finite setting (exact game) whereas the determination of the NE is performed in the asymptotic regime (approximated game). It turns out that arge system approximates of ergodic transmission rates have the same properties as their exact counterparts, as shown recenty by [23], which therefore ensures the existence and uniqueness of the NE in the approximated game. Nash Equiibrium efficiency. In order to measure the efficiency of the decentraized networ w.r.t. its centraized counterpart we introduce the foowing quantity: SRE = RNE sum C sum, (7)

11 where SRE stands for sum-rate efficiency; the quantityrsum NE represents the sum-rate of the decentraized networ at the Nash equiibrium, which is achieved for certain choices of coding and decoding strategies; the quantity C sum corresponds to the sum-capacity of the centraized networ, which is reached ony if the optimum coding and decoding schemes are nown. Note that this is the case for the MAC but not for other channes ie the interference channe. Obviousy, the efficiency measure we introduce here is strongy connected to the price of anarchy [24] (POA). The difference between SRE and POA is subte. In our context, information theory provides us with fundamenta physica imits on the socia wefare (networ sum-capacity) whie in genera no such upper bound is avaiabe. In our case, the sum-capacity is given by: K C sum = max I+ρ H Ω H H, (8) with (Ω,...,Ω K) A (C) Eog = A (C) = { (Ω,...,Ω K ) K,Ω 0,Ω = Ω H,Tr(Ω ) n t P }. (9) In genera, it is not easy to find a cosed-form expression of the SRE. This is why we wi respectivey anayze the SRE in the regimes of high and ow signa-to-noise ratio (SNR), and for intermediate regimes simuations wi compete our anaysis. It turns out that the SRE tends to in the two mentioned extreme regimes, which is the purpose of what foows. In the high SNR regime, where ρ, we observe from (4) that δ (π) (j) is easy to chec that by setting the derivatives of L w.r.t. P (s) poicy at the NE is the uniform power aocation P (π),ne γ (π) (j). Under this condition, it (j) to zero, we obtain that the power aocation = P I, regardess the reaization of the coordination signa S. Furthermore, in the high SNR regime, the sum-capacity is achieved by the uniform power aocation. Thus, we obtain that the gap between the NE achievabe sum-rate and the sum-capacity is optima, SRE = for any distribution of S. In the ow SNR regime, whereρ 0, from (4) we obtain thatδ (π) (j) 0 and thatγ (π) n r (j) = σ (N (π) (i,j). +)nt i= By approximating n(+x) x when x <<, the power aocations poicies at the NE are the soutions of the foowing inear programs: {P (π) max (j)} j nt { } n t p π P (π) n r (j) σ (i,j) j= i= n t, (20) s.t. (j) P n t P (π) j=

12 2 given by: n r p π P (π),ne n t P if j = arg max σ (i,m) (j) = m n t i= 0 otherwise. (2) The optima power aocation that achieves the sum-capacity is equa to the equiibrium power aocation, P = p π P (π),ne (j) Thus, the achievabe sum-rate at the NE is equa to the centraized upper bound and thus SRE = for any distribution of S. In concusion, when either the ow or high SNR regime is assumed, the sum-capacity of the fast fading MAC is achieved at the NE athough a sub-optimum coordination mechanism is assumed and aso regardess of the distribution of the coordination channe. IV. SINGLE USER DECODING In this section the coordination signa is deterministic (namey Pr[S = s] = δ(s), δ being the Kronecer symbo) and therefore the amount of downin signaing the BS needs in order to indicate to the MSs that it is using SUD can be made arbitrary sma (by etting the frequency at which the reaizations of the coordination signa are drawn tend to zero). In this framewor, each user has to optimize ony one precoding matrix. Indeed, [ ] the strategy of user K, consists in choosing the best precoding matrix Q (0) = E X (0) X(0)H, in the sense of his utiity function obtained with SUD: u SUD (Q (0),Q(0) ) = Eog I+ρH Q (0). The strategy set of user becomes A SUD = HH +ρ H Q (0) H H Eog I+ρ H Q (0) H H (22) { Q (0) 0,Q (0) = Q (0),H,Tr(Q (0) ) n tp }. (23) It turns out that the equiibrium anaysis in the game with SUD can be, to a arge extent, deduced from the game with SIC. For this reason, we wi not detai the corresponding proofs. The existence and uniqueness issues are given in the foowing theorem. Theorem 6: [Existence and uniqueness of an NE] The space power aocation game described by: the set of payers K; the sets of actions A SUD unique Nash equiibrium. and the payoff functions u SUD (Q (0),Q(0) ) given in (22), has a To prove the existence of a Nash equiibrium we aso expoit Theorem and the four necessary conditions on the utiity functions and strategy sets can be verified using the same toos as described in Appendix A. Uniqueness of the Nash equiibrium. Here we can speciaize Theorem 4, which is the matrix extension of Theorem 2. When the strategies sets are not sets of pairs of matrices but ony sets of matrices, the diagonay strict

13 3 concavity condition in (6) can be written as foows. For a Q (0),Q (0) A SUD such that (Q (0),...,Q (0) K ) (Q (0),...,Q (0) K ): C = K Tr = [ {(Q (0) Q (0) ) Now we can evauate C and obtain that: K C = E Tr [ρh (Q (0) = ( K I+ρ = Q (0) u (Q (0),Q (0) Q (0) )H H ) H Q (0) H H = ETr{(B B )[(B ) (B ) ]}, = E[F 0 (H)] which is positive for any B = I+ K H Q (0) H H, = I+ B = to prove that for any (Q (0),...,Q (0) K ) (Q(0),...,Q (0) K ) Q (0) ] ( I+ρ K = ]} u (Q (0),Q (0) ). (24) H Q (0) H H ) (25) K H Q (0) H H from (8) for K = 2. We need = ) we have C > 0. Remar: Assuming that ran(h H H) = Kn t and Kn t n r + K, then Q Q impies that B B. This means that for any channe reaization we have F 0 (H) > 0 which impies directy that C > 0. For the genera proof, we define the foowing sets: { B H (Q (0),Q (0) ) = H D H } K = H (0) (Q (0) Q )H H { = 0 B H (Q (0),Q (0) ) = H D H } (26) K = H (0) (Q (0) Q )H H 0 We now that: C = E[F 0 (H)] = F 0 (H)L(H)dH D H = F 0 (H)L(H)dH+ F 0 (H)L(H)dH B H(Q (0),Q (0) ) B H(Q (0),Q (0) ) = F 0 (H)L(H)dH B H(Q (0),Q (0) ) (27) The second equaity foows since D H = B H (Q (0),Q (0) ) B H (Q (0),Q (0) ). The third equaity foows because F 0 (H) = 0 for a H B H (Q (0),Q (0) ) from Lemma. We aso now that F 0 (H) > 0 for a H B H (Q (0),Q (0) ). It suffices to prove that BH (Q (0),Q (0) ) is a subset of non-zero Lebesgue measure to impy that C > 0. Here as we, the existence of the compact set can be proved (simiary to the proof for the SIC decoding technique).

14 4 Determination of the Nash equiibrium. As for the optima eigenvectors of the covariance matrices, we foow the same ines as in Appendix C. In this case aso there is no oss of optimaity by choosing the covariance matrices Q (0) = W P (0) WH, where W is the same unitary matrix as in (2) and P is the diagona matrix containing the eigenvaues of Q (0). Here aso we further expoit the asymptotic resuts for the MIMO channe given in [20] [2]. The approximated utiity for user is: ũ SUD (P (0),P(0) ) = K n t og n 2 (+KρP (0) (j)γ (j))+ r = j= n r og n 2 + K n t σ (i,j)δ (j) r Kn i= t = j= K n t γ (j)δ (j)og 2 e n r = j= n r n r i= n r n t og 2 (+(K )ρp (0) (j)φ (j)) j= n r n t og 2 + σ (i,j)ψ (j) + (K )n t n r j= φ (j)ψ (j)og 2 e j= where the parameters γ (j) and δ (j) j {,...,n t }, {,2} are soution of: γ (j) = j {,...,n t }, K : n r σ (i,j) Kn t K n t i= + σ (i,m)δ (m) Kn t = m= KρP (0) δ (j) = (j) +KρP (0) (j)γ (j). and φ (j), ψ (j), j {,...,n t } are the unique soutions of the foowing system: φ (j) = ψ (j) = j {,...,n t }, K\{} : n r σ (i,j) (K )n t n t i= + (K )n t σ r (i,m)ψ r (m) (K )ρp (0) (j) +(K )ρp (0) (j)φ (j). r m= (28) (29) (30)

15 5 The corresponding water-fiing soution is: [ P (0),NE (j) = n2n r λ ] +, (3) Kργ (j) [ n t whereλ 0 is the Lagrangian mutipier tuned in order to meet the power constraint: n t P. j= n2n r λ ] + = Kργ (j) In what the efficiency of the NE point is concerned, we aready now that the SUD decoding technique is sub-optima in the centraized case (SUD does aow the networ to operate at an arbitrary point of the centraized MAC capacity region) and it is impossibe to reach the sum-capacity C sum even if the high and ow SNR regime are assumed. V. SIMULATION RESULTS In what foows, we assume the regime of arge numbers of antennas. From [9], [20], [2], we now that the approximates of the ergodic achievabe rates in the asymptotic regime are accurate even for reativey sma number of antennas. For the channe matrices, we assume the Kronecer mode H = R /2 Θ T /2 mentioned in Sec. II, where the receive and transmit correation matricesr, T foow an exponentia profie characterized by the correation coefficients (see e.g., [25], [26]) r = [r,r 2 ] and t = [t,t 2 ] such that R (i,j) = r i j, T (i,j) = t i j. By assuming that the receive antenna is a uniform inear array (ULA) and nowing that, when the dimensions of Toepitz matrices increase they can be approximated by circuar matrices we obtain that a the receive correation matrices R can be diagonaized in the same vector basis (i.e., the Fourier basis). Thus the considered mode is incuded in the UIU mode that we studied where V = V. Fair SIC decoding versus SUD decoding. First we compare the resuts of the genera space-time PA game considered in Sec. III, where SIC decoding is used at the receiver, and the game described in Sec. IV, where SUD decoding is used. Fig. depicts the achievabe sum-rate at the equiibrium as a function of the transmit power P = P 2 = P, for the scenario n r = n t = 0, r = [0.5,0.2], t = [0.5,0.2], ρ = 3dB. In order to have a fair comparison we assume that p = 2 (on average each user is decoded second haf of the time when SIC is assumed). We observe that, even in this scenario, which was thought to be a bad one in terms of sub-optimaity, the sum-rate obtained with the first game is very cose to the sum-capacity upper bound. Aso, the sum-rate reached when the BS uses SUD is ceary much ower than the sum-rate obtained by using SIC. SIC decoding, comparison between the joint space-time PA and the specia cases of spatia PA and tempora PA. Now we want to compare the resuts of the genera space-time PA with the two particuar cases that were

16 6 studied in [4]: the spatia PA, where the users are forced to aocate their power uniformy over time (regardess of their decoding ran) but are free to aocate their power over the transmit antennas; the tempora PA, where the users are forced to aocate their power uniformy over their antennas but they can adjust their power as a function of the decoding ran at the receiver. Fig. 2 represents the sum-rate efficiency as a function of the coordination signa distribution parameter p [0,] when n r = n t = 0, r = [0.3,0], t = [0.5,0.2], ρ = 4dB, P = 5, P 2 = 50. We observe that the three types of power aocation poicies perform very cose to the upper bound. What is most interesting is the fact that the performance of the networ at the equiibrium is better by using a purey spatia PA instead of the most genera space-time PA. This has been confirmed by many other simuations and iustrates a Braess paradox: athough the sets of strategies for the space-time case incude those of the purey spatia case, the performance obtained at the NE are not better in the space-time case. SIC decoding, spatia PA, achievabe rate region. In Fig. 3, we observe that the rate region achieved at the NE of the space PA as a function of the distribution of the coordination signa p for the scenario n r = n t = 0, r = [0.4,0.2], t = [0.6,0.3], ρ = 3dB, P = 5, P 2 = 50. It is quite remarabe that in arge MIMO MACs, the capacity region comprises a fu cooperation segment just ie the SISO MACs. The coordination signa precisey aows one to move aong the corresponding ine. This shows the reevance of arge systems in decentraized networs since they aow to determine the capacity region of certain systems whereas it is unnown in the finite setting. Furthermore, they induce an averaging effect, which maes the users behavior predictabe. VI. CONCLUSIONS Interestingy, the existence and uniqueness of the Nash equiibrium can be proven in mutipe access channes with muti-antenna terminas for a genera propagation channe mode (namey the unitary-invariant-unitary mode) and the most genera case of space-time power aocation schemes. In particuar, the uniqueness proof requires a matrix generaization of the second theorem of Rosen [7] and proving a trace inequaity [28]. For a the types of power aocation poicies (purey tempora PA, purey spatia PA, space-time PA), the sum-rate efficiency of the decentraized networ is cose to one when SIC is assumed and the networ is coordinated by the proposed suboptimum coordination mechanism. Quite surprisingy, the space-time power aocation performs a itte worse than its purey spatia counterpart, which puts in evidence a Braess paradox in the types of wireess networs under consideration. One of the interesting extensions of this wor woud be to anayze the impact of a non-perfect SIC on the PA probem. Indeed, the effect of propagation errors coud then be assessed (which does not exist with SUD).

17 7 APPENDIX A A. Concavity of the utiity functions u SIC Let us focus on user K. We want to prove that u SIC (Q,Q ) is concave w.r.t. Q A SIC. We observe that the term R (π) (Q(π),Q(π) ) in (3) depends ony on Q(π) and Q (π) and not on the covariance matrices Q(τ), Q (τ) for any other possibe decoding rue τ P K \{π}. Thus, in order to prove that u SIC (Q,Q ) is stricty concave w.r.t. to Q = (Q (π) ), it suffices to prove that R (π) (Q(π) π P K.,Q(π) To this end, we study the concavity of the function f(λ) = R (π) (λq(π) [0,] for any pair of matrices (Q (π) 2 f λ (λ) = ETr ρ 2 H H 2 H H,Q (π) ). The second derivative of f is equa to: I+ρH Q (π) H H +ρλh Q (π) HH +ρ I+ρH Q (π) H H +ρλh Q (π) HH +ρ = ETr[A Q (π) A Q(π) ] with A = ρ 2 H H I+ρH Q (π) H H +ρλh Q (π) HH +ρ to be a Hermitian positive definite matrix, Q (π) = Q (π) with B = A /2 Q (π) A/2. K (π) K (π) ) is concave w.r.t. Q(π) for a +( λ)q (π) ) over the interva K (π) H Q (π) H H H Q (π) H H 2 f λ (λ) = ETr[A /2 Q (π) 2 A/2 A /2 Q (π) A/2 ] = ETr[BB H ] < 0 H Q (π) H H H Q (π) H Q (π) H, which can be proven Q (π) aso a Hermitian matrix, and ρ = σ 2.,, B. Continuity of the utiity functions u SIC Considering the Leibniz formua, the determinant of a matrix can be expressed as a weighted sum of products of its entries. Knowing that the product and the sum of continuous functions are continuous, we concude that the determinant function is continuous. Aso, it is we nown that the ogarithmic function is a continuous function. Thus, for any π P K, the function R (π) (Q(π) continuous functions which is aso continuous w.r.t. (Q (π) is continuous w.r.t. (Q,Q ).,Q(π),Q(π) ) is nothing ese but the composition of two ). This suffices to prove that usic (Q,Q )

18 8 C. Convexity of the strategy sets A SIC In order to prove that the set A SIC A SIC A SIC, we have: is convex, we need to verify that, for any two matrices (Q,Q ) αq +( α)q A SIC, for a α 0. For any Q,Q A(SIC), the matrices Q (π) are Hermitian which impies that αq (π) +( α)q (π) aso Hermitian matrices, for a π P K. Furthermore, for any Q,Q ASIC, we have that Q (π), Q (π) that αq (π) +( α)q (π) are aso non-negative matrices, for a π P K. Finay, nowing that the trace is a inear appication we have that: ) p π Tr (αq (π) +( α)q (π) = π P = α p π Tr(Q (π) )+( α) p π Tr(Q (π) π P π P αn t P +( α)n t P = n t P. are are non-negative matrices which impies ) Thus αq +( α)q ASIC and the set is convex. D. Compactness of the strategy sets A SIC To prove that the strategy sets are compact sets we use the fact that, in finite dimension spaces, a cosed and bounded set is compact. First et us prove that A SIC is a cosed set. We define the function g : A SIC f(q ) = p π Tr(Q (π) ). [0,n t P ], with We see that g( ) is a continuous function and that its image is a compact and thus cosed set. Knowing that the continuous inverse image of a cosed set is cosed, we concude that A SIC Now we want to prove that the set A SIC the foowing norm Q = is cosed. is a bounded set. We associate to the tupe of matrices (Q (π) ) Q (π) 2 2 where. 2 is is the spectra norm of a matrix. Q (π) 2 = max{λ Q (π)h Q (i)} n (π) i=.

19 9 Since for a Q A SIC, Q (π) is a non-negative, Hermitian matrix we have that: max{λ (π) Q (i)} n i= Tr(Q(π) ), and thus: Q (π) 2 = max{λ Q (π)2 (i)} n i= max{λ = (π) Q (i) 2 } n i=. In concusion the associated norm Q. APPENDIX B We suppose that there exist two different equiibrium strategy profies: ( Q, Q ) A SIC ( Q, Q ) A SIC A SIC and A SIC, such that ( Q, Q ) ( Q, Q ). Then the condition given in the theorem, C > 0 is met for the particuar choice of (Q,Q ) = ( Q, Q ) and (Q,Q ) = ( Q, Q ). By the definition of the Nash Equiibrium, the strategies Q, K, are the soutions of the foowing maximization probems: max Q A SIC u (Q, Q ). Thus, Q satisfy the foowing Kuhn-Tucer optimaity conditions: ) Q A SIC, which means that: Q (π) = ( Q (π) )H 0 p π Tr( Q (π) ) n tp,, π P K 2) There exist λ 0, and the foowing Hermitian non-negative matrices of ran, such that: [ ] λ p π Tr( Q (π) ) n tp = 0 Tr( Φ (π) Q (π) ) = 0, π P K, Φ (π), for a π P K, 3) π P K : (π) Q u ( Q, Q (π) ) = p π λ I Φ,

20 20 Having assumed that ( Q, Q ) is aso a Nash Equiibrium, Q, with K are the soution of: max Q A SIC u (Q, Q ), and thus Q satisfy the foowing Kuhn-Tucer optimaity conditions: 4) Q A SIC, which means that: Q (π) = ( Q (π) )H 0 p π Tr( Q (π) ) n tp,, π P K 5) There exist λ 0, K and the foowing non-negative, Hermitian matrices of ran, π P K such that: [ ] λ p π Tr( Q (π) ) n tp = 0 Tr( Φ (π) Q (π) ) = 0, π P K, (π) Φ, for a 6) π P K : Q (π) u ( Q, Q (π) ) = p π λ I Φ Using the third and the sixth optimaity conditions, the condition given in (6) becomes: C = K = Tr( Q (π) K = 0. { p π λ Tr( Q (π) )+p πˆλ Tr( Q (π) ) p π λ Tr( Q (π) ) p πˆλ Tr( Q (π) ) Φ (π) { λ [ } (π) ) Tr( Q Φ (π) (π) )+Tr( Q Φ (π) (π) )+Tr( Q Φ (π) ] [ ) ]} p π Tr( Q (π) ) n tp + ˆλ p π Tr( Q (π) ) n tp From the other four K-T conditions, we obtain that a the terms on the right are negative and thus C 0. But this contradicts the diagonay strict concavity condition and so the Nash Equiibrium is unique.. APPENDIX C We want to prove that there is no optimaity oss when restricting the search for the optima covariance matrices to Q A SIC such that Q (π) = W P (π) WH, for a π P K. Let us consider user K. We have

21 2 that: arg max Q A SIC = arg max Q A SIC = arg max Q A SIC = arg max Q A SIC = arg max Q A SIC where we denoted with X (π) u (Q,Q ) p π Eog 2 I+ρH Q (π) HH +ρ H Q (π) H H K (π) p π Eog 2 I+ρV H W H Q(π) W H H VH +ρ V H W H Q(π) W HH V H, K (π) p π Eog 2 I+ρ H W H Q(π) W H H +ρ H W H Q(π) W HH K (π) Eog 2 I+ρ H X (π) H H +ρ H W H Q(π) W HH π K (π) K (π) (32) W H Q(π) W. Knowing that the utiity function is concave w.r.t. the new defined matrices X (π), and the channe matrix H has independent entries, we can directy appy the resuts given in [22] to prove that annuing the non-diagona entries of X (π) can ony increase the vaues of the functions Eog 2 I+ρ H X (π) H H +ρ H W H Q(π) W HH. In concusion the optima matrices X(π) are diagona, that we wi denote with P (π) Q (π) = W P (π) WH. K (π). The spectra decomposition of the optima covariance matrices are: REFERENCES [] S. A. Grandhi, R. Vijayan and D. J. Goodman, Distributed agorithm for power contro in ceuar radio systems, Proc. of Annua Aerton Conf. on Comm. Contro and Computing, Sep [2] S. A. Grandhi, R. Vijayan and D. J. Goodman, Distributed power contro in ceuar radio systems, IEEE Trans. on Comm., Vo. 42, No. 234, pp , Feb/Mar/Apr 994. [3] H. Ji and C.-Y. Huang, Non-cooperative upin power contro in ceuar radio systems, Wireess Networs, Vo. 4, No. 3, pp , 998. [4] S.-J. Oh, T. L. Osen and K. M. Wasserman, Distributed power contro and spreading gain aocation in CDMA data networs, IEEE Proc. of INFOCOM, Vo. 2, March 2000, pp [5] D. J. Goodman and N. B. Mandayam, Power Contro for Wireess Data, IEEE Person. Comm., Vo. 7, No. 2, pp , Apri [6] F. Meshati, M. Chiang, H. V. Poor and S. C. Schwartz, A game-theoretic approach to energy-efficient power contro in muti-carrier CDMA systems, IEEE Journa on Seected Areas in Communications, Vo. 24, No. 6, pp. 5 29, June [7] W. Yu, G. Ginis and J. M. Cioffi, Distributed mutiuser power contro for digita subscriber ines, IEEE Journa of Seected Areas in Communications, Vo. 20, No. 5, June 2002, pp

22 22 7 P =P P 2 =0P n r =n t =0 r=[ ] t=[ ] ρ=3db 6 5 Achievabe sum rate Sum Capacity Fair SIC decoding (p=/2) SUD decoding P Fig.. Fair SIC (joint space-time power aocation) vs. SUD decoding. Achievabe networ sum-rate versus the avaiabe transmit power P for p =, nr = nt = 0, r = [0.5,0.2], t = [0.5,0.2], ρ = 3dB. The fair SIC performs much coser to the sum-capacity 2 upper bound than SUD. [8] L. Lai and H. E Gama, The Water-Fiing Game in Fading Mutipe-Access Channes, IEEE Trans. on Information Theory, Vo. 54, No. 5, pp , May [9] S. Lasauce, A. Suarez, M. Debbah and L. Cottateucci, Power aocation game for fading MIMO mutipe access channes with antenna correation, in the ICST/ACM proc. of the Internationa Conf. on Game Theory in Comm. Networs (Gamecomm), Nantes, France, Oct [0] G. Arsan, M. F. Demiro and Y. Song, Equiibrium Efficiency Improvement in MIMO Interference Systems: a Decentraized Stream Contro Approach, IEEE Trans. on Wireess Communications, Vo. 6, No. 8, pp , Aug [] G. Scutari, D. P. Paomar and S. Barbarossa, Competitive Design of Mutiuser MIMO Systems Based on Game Theory: A Unified View, IEEE Journa of Seected Areas in Communications, Vo. 26, No. 7, pp , Sept [2] A. Careia, Interference channes, IEEE Trans. on Inform. Theory, Vo. 24, No., pp , 978. [3] D. P. Paomar, J. M. Cioffi, M. A. Lagunas, Uniform Power Aocation in MIMO Channes: A Game Theoretic Approach, IEEE Trans. on Information Theory, Vo. 49, No. 7, pp , Juy [4] E. V. Bemega, S. Lasauce and M. Debbah, Power contro in distributed mutipe access channes with coordination, Internat. Worshop on Wireess Networs: Communication, Cooperation and Competition (WNC3), Apri [5] A. D. Wyner, Recent resuts in Shannon theory, IEEE Trans. on Inform. Theory, Vo. 20, pp. 2 0, Jan. 974.

23 n r =n t =0 r=[0.3 0] t=[ ] P=5 P2=50 ρ=4 db Sum capacity Spatia PA Joint Space Time PA Tempora PA 99 Sum rate efficiency η[%] p Fig. 2. SIC decoding, comparaison between the joint space-time PA and the two specia cases: the space PA and tempora PA. Sum-rate efficiency versus the distribution of the coordination signa p [0,] for n r = n t = 0, r = [0.3,0], t = [0.5,0.2], ρ = 4dB, P = 5, P 2 = 50. The spatia PA outperforms the joint space-time PA (Braess paradox). [6] T. Cover, Some advances in broadcast channes, in Advances in Communication Systems, Vo. 4, Academic Press, 975. [7] J. Rosen, Existence and uniqueness of equiibrium points for concave n-person games, Econometrica, Vo. 33, pp , 965. [8] E. Bigieri, G. Taricco and A. Tuino, How far is infinity? Using asymptotic anayses in mutipe-antennas systems, Proc. of the Int Symposium on Software Testing and Anaysis (ISSTA), Vo., pp. 6, [9] J. Dumont, P. Loubaton and S. Lasauce, On the capacity achieving transmit covariance matrices of MIMO correated Rician channes: a arge system approach, IEEE Proc. of Gobecom Technica Conference, Nov/Dec [20] A. Tuino and S. Verdu, Random Matrices and Wireess Communications, Foundations and trends in comm. and inform. theory, NOW, The Essence of Knowedge, [2] A. Tuino and S. Verdu, Impact of antenna correation on the capacity of muti-antenna channes, IEEE Trans. on Inform. Theory, Vo. 5, No. 7, pp , Juy [22] A. Tuino, A. Lozano and S. Verdu, Capacity-achieving input covariance for singe-user muti-antenna channes, IEEE Trans. on Wireess Communications, Vo. 5, No. 3, pp , March [23] J. Dumont, W. Hachem, S. Lasauce, P. Loubaton and J. Najim, On the capacity achieving covariance matrix of Rician MIMO channes: an asymptotic approach, IEEE Trans. on Inform. Theory, submitted. [24] T. Roughgarden and E. Tardos, How bad is sefish routing?, Journa of the ACM (JACM), Vo. 49, No. 2, pp , March 2002.

24 Achievabe rate region (n r =n t =0 r=[ ] t=[ ] P=5 P2=50 ρ=3 db) Sum capacity SIC decoding SUD decoding R 2 (p) R (p) Fig. 3. SIC decoding, space PA. The achievabe rate region at the NE versus the distribution of the coordination signa p [0,] for n r = n t = 0, r = [0.4,0.2], t = [0.6,0.3], ρ = 3dB, P = 5, P 2 = 50. Varying p aows to move aong a segment cose to the sum-capacity, simiar to SISO MAC. [25] M. Chiani, M. Z. Win, A. Zanea, On the capacity of spatiay correated MIMO Rayeiggh-Fading Channes, IEEE Trans. on Inform. Theory, Vo. 49, No. 0, pp , Juy [26] A. Supch, D. Seethaer, F. Hawatsch, Free probabiity based capacity cacuation for MIMO channes with transmit or receive correation, Internationa Conference on Wireess Networs, Communications and Mobie Computing, June [27] E. V. Bemega, S. Lasauce and M. Debbah, A trace inequaity for positive definite matrices, Journa of Inequaities in Pure and Appied Mathematics (JIPAM), Vo. 0, No., pp. -4, [28] E. V. Bemega, M. Jungers and S. Lasauce, A generaization of a trace inequaity for positive definite matrices., The Austraian Journa of Mathematica Anaysis and Appications (AJMAA), to appear, 200. [29] K. M. Abadir AND J. R. Magnus, Matrix Agebra, Cambridge University Press, New Yor, USA, [30] R. A. Horn AND C. R. Johnson, Matrix Anaysis, Cambridge University Press, New Yor, USA, 99.

Maximizing Sum Rate and Minimizing MSE on Multiuser Downlink: Optimality, Fast Algorithms and Equivalence via Max-min SIR

Maximizing Sum Rate and Minimizing MSE on Multiuser Downlink: Optimality, Fast Algorithms and Equivalence via Max-min SIR 1 Maximizing Sum Rate and Minimizing MSE on Mutiuser Downink: Optimaity, Fast Agorithms and Equivaence via Max-min SIR Chee Wei Tan 1,2, Mung Chiang 2 and R. Srikant 3 1 Caifornia Institute of Technoogy,

More information

Source and Relay Matrices Optimization for Multiuser Multi-Hop MIMO Relay Systems

Source and Relay Matrices Optimization for Multiuser Multi-Hop MIMO Relay Systems Source and Reay Matrices Optimization for Mutiuser Muti-Hop MIMO Reay Systems Yue Rong Department of Eectrica and Computer Engineering, Curtin University, Bentey, WA 6102, Austraia Abstract In this paper,

More information

A Brief Introduction to Markov Chains and Hidden Markov Models

A Brief Introduction to Markov Chains and Hidden Markov Models A Brief Introduction to Markov Chains and Hidden Markov Modes Aen B MacKenzie Notes for December 1, 3, &8, 2015 Discrete-Time Markov Chains You may reca that when we first introduced random processes,

More information

Power Allocation Games for. MIMO Multiple Access Channels with Coordination

Power Allocation Games for. MIMO Multiple Access Channels with Coordination Power Allocation Games for MIMO Multiple Access Channels with Coordination Elena-Veronica Belmega, Student Member, IEEE, Samson Lasaulce, Member, IEEE, and Merouane Debbah, Senior Member, IEEE Abstract

More information

T.C. Banwell, S. Galli. {bct, Telcordia Technologies, Inc., 445 South Street, Morristown, NJ 07960, USA

T.C. Banwell, S. Galli. {bct, Telcordia Technologies, Inc., 445 South Street, Morristown, NJ 07960, USA ON THE SYMMETRY OF THE POWER INE CHANNE T.C. Banwe, S. Gai {bct, sgai}@research.tecordia.com Tecordia Technoogies, Inc., 445 South Street, Morristown, NJ 07960, USA Abstract The indoor power ine network

More information

Lecture Note 3: Stationary Iterative Methods

Lecture Note 3: Stationary Iterative Methods MATH 5330: Computationa Methods of Linear Agebra Lecture Note 3: Stationary Iterative Methods Xianyi Zeng Department of Mathematica Sciences, UTEP Stationary Iterative Methods The Gaussian eimination (or

More information

LINEAR DETECTORS FOR MULTI-USER MIMO SYSTEMS WITH CORRELATED SPATIAL DIVERSITY

LINEAR DETECTORS FOR MULTI-USER MIMO SYSTEMS WITH CORRELATED SPATIAL DIVERSITY LINEAR DETECTORS FOR MULTI-USER MIMO SYSTEMS WITH CORRELATED SPATIAL DIVERSITY Laura Cottateucci, Raf R. Müer, and Mérouane Debbah Ist. of Teecommunications Research Dep. of Eectronics and Teecommunications

More information

Reliability: Theory & Applications No.3, September 2006

Reliability: Theory & Applications No.3, September 2006 REDUNDANCY AND RENEWAL OF SERVERS IN OPENED QUEUING NETWORKS G. Sh. Tsitsiashvii M.A. Osipova Vadivosto, Russia 1 An opened queuing networ with a redundancy and a renewa of servers is considered. To cacuate

More information

Scalable Spectrum Allocation for Large Networks Based on Sparse Optimization

Scalable Spectrum Allocation for Large Networks Based on Sparse Optimization Scaabe Spectrum ocation for Large Networks ased on Sparse Optimization innan Zhuang Modem R&D Lab Samsung Semiconductor, Inc. San Diego, C Dongning Guo, Ermin Wei, and Michae L. Honig Department of Eectrica

More information

MC-CDMA CDMA Systems. Introduction. Ivan Cosovic. Stefan Kaiser. IEEE Communication Theory Workshop 2005 Park City, USA, June 15, 2005

MC-CDMA CDMA Systems. Introduction. Ivan Cosovic. Stefan Kaiser. IEEE Communication Theory Workshop 2005 Park City, USA, June 15, 2005 On the Adaptivity in Down- and Upink MC- Systems Ivan Cosovic German Aerospace Center (DLR) Institute of Comm. and Navigation Oberpfaffenhofen, Germany Stefan Kaiser DoCoMo Euro-Labs Wireess Soution Laboratory

More information

A. Distribution of the test statistic

A. Distribution of the test statistic A. Distribution of the test statistic In the sequentia test, we first compute the test statistic from a mini-batch of size m. If a decision cannot be made with this statistic, we keep increasing the mini-batch

More information

Transmit Antenna Selection for Physical-Layer Network Coding Based on Euclidean Distance

Transmit Antenna Selection for Physical-Layer Network Coding Based on Euclidean Distance Transmit ntenna Seection for Physica-Layer Networ Coding ased on Eucidean Distance 1 arxiv:179.445v1 [cs.it] 13 Sep 17 Vaibhav Kumar, arry Cardiff, and Mar F. Fanagan Schoo of Eectrica and Eectronic Engineering,

More information

A Simple and Efficient Algorithm of 3-D Single-Source Localization with Uniform Cross Array Bing Xue 1 2 a) * Guangyou Fang 1 2 b and Yicai Ji 1 2 c)

A Simple and Efficient Algorithm of 3-D Single-Source Localization with Uniform Cross Array Bing Xue 1 2 a) * Guangyou Fang 1 2 b and Yicai Ji 1 2 c) A Simpe Efficient Agorithm of 3-D Singe-Source Locaization with Uniform Cross Array Bing Xue a * Guangyou Fang b Yicai Ji c Key Laboratory of Eectromagnetic Radiation Sensing Technoogy, Institute of Eectronics,

More information

Problem set 6 The Perron Frobenius theorem.

Problem set 6 The Perron Frobenius theorem. Probem set 6 The Perron Frobenius theorem. Math 22a4 Oct 2 204, Due Oct.28 In a future probem set I want to discuss some criteria which aow us to concude that that the ground state of a sef-adjoint operator

More information

Power Control and Transmission Scheduling for Network Utility Maximization in Wireless Networks

Power Control and Transmission Scheduling for Network Utility Maximization in Wireless Networks ower Contro and Transmission Scheduing for Network Utiity Maximization in Wireess Networks Min Cao, Vivek Raghunathan, Stephen Hany, Vinod Sharma and. R. Kumar Abstract We consider a joint power contro

More information

Asynchronous Control for Coupled Markov Decision Systems

Asynchronous Control for Coupled Markov Decision Systems INFORMATION THEORY WORKSHOP (ITW) 22 Asynchronous Contro for Couped Marov Decision Systems Michae J. Neey University of Southern Caifornia Abstract This paper considers optima contro for a coection of

More information

LEARNING DISTRIBUTED POWER ALLOCATION POLICIES IN MIMO CHANNELS

LEARNING DISTRIBUTED POWER ALLOCATION POLICIES IN MIMO CHANNELS LEARNING DISTRIBUTED POWER ALLOCATION POLICIES IN MIMO CHANNELS Elena Veronica Belmega, Samson Lasaulce, Mérouane Debbah and Are Hjørungnes LSS (joint lab of CNRS, SUPELEC, Univ. Paris-Sud 11) Gif-sur-Yvette

More information

Iterative Decoding Performance Bounds for LDPC Codes on Noisy Channels

Iterative Decoding Performance Bounds for LDPC Codes on Noisy Channels Iterative Decoding Performance Bounds for LDPC Codes on Noisy Channes arxiv:cs/060700v1 [cs.it] 6 Ju 006 Chun-Hao Hsu and Achieas Anastasopouos Eectrica Engineering and Computer Science Department University

More information

Polite Water-filling for the Boundary of the Capacity/Achievable Regions of MIMO MAC/BC/Interference Networks

Polite Water-filling for the Boundary of the Capacity/Achievable Regions of MIMO MAC/BC/Interference Networks 2011 IEEE Internationa Symposium on Information Theory Proceedings Poite Water-fiing for the Boundary of the Capacity/Achievabe Regions of MIMO MAC/BC/Interference Networks An iu 1, Youjian iu 2, Haige

More information

Intuitionistic Fuzzy Optimization Technique for Nash Equilibrium Solution of Multi-objective Bi-Matrix Games

Intuitionistic Fuzzy Optimization Technique for Nash Equilibrium Solution of Multi-objective Bi-Matrix Games Journa of Uncertain Systems Vo.5, No.4, pp.27-285, 20 Onine at: www.jus.org.u Intuitionistic Fuzzy Optimization Technique for Nash Equiibrium Soution of Muti-objective Bi-Matri Games Prasun Kumar Naya,,

More information

LOW-COMPLEXITY LINEAR PRECODING FOR MULTI-CELL MASSIVE MIMO SYSTEMS

LOW-COMPLEXITY LINEAR PRECODING FOR MULTI-CELL MASSIVE MIMO SYSTEMS LOW-COMPLEXITY LINEAR PRECODING FOR MULTI-CELL MASSIVE MIMO SYSTEMS Aba ammoun, Axe Müer, Emi Björnson,3, and Mérouane Debbah ing Abduah University of Science and Technoogy (AUST, Saudi Arabia Acate-Lucent

More information

Sum Capacity and TSC Bounds in Collaborative Multi-Base Wireless Systems

Sum Capacity and TSC Bounds in Collaborative Multi-Base Wireless Systems IEEE TRANSACTIONS ON INFORMATION THEORY, VOL X, NO X, DECEMBER 004 1 Sum Capacity and TSC Bounds in Coaborative Muti-Base Wireess Systems Otiia Popescu, Student Member, IEEE, and Christopher Rose, Member,

More information

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents MARKOV CHAINS AND MARKOV DECISION THEORY ARINDRIMA DATTA Abstract. In this paper, we begin with a forma introduction to probabiity and expain the concept of random variabes and stochastic processes. After

More information

The Streaming-DMT of Fading Channels

The Streaming-DMT of Fading Channels The Streaming-DMT of Fading Channes Ashish Khisti Member, IEEE, and Star C. Draper Member, IEEE arxiv:30.80v3 cs.it] Aug 04 Abstract We consider the sequentia transmission of a stream of messages over

More information

Distributed average consensus: Beyond the realm of linearity

Distributed average consensus: Beyond the realm of linearity Distributed average consensus: Beyond the ream of inearity Usman A. Khan, Soummya Kar, and José M. F. Moura Department of Eectrica and Computer Engineering Carnegie Meon University 5 Forbes Ave, Pittsburgh,

More information

Duality, Polite Water-filling, and Optimization for MIMO B-MAC Interference Networks and itree Networks

Duality, Polite Water-filling, and Optimization for MIMO B-MAC Interference Networks and itree Networks Duaity, Poite Water-fiing, and Optimization for MIMO B-MAC Interference Networks and itree Networks 1 An Liu, Youjian Eugene) Liu, Haige Xiang, Wu Luo arxiv:1004.2484v3 [cs.it] 4 Feb 2014 Abstract This

More information

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES Separation of variabes is a method to sove certain PDEs which have a warped product structure. First, on R n, a inear PDE of order m is

More information

Optimality of Gaussian Fronthaul Compression for Uplink MIMO Cloud Radio Access Networks

Optimality of Gaussian Fronthaul Compression for Uplink MIMO Cloud Radio Access Networks Optimaity of Gaussian Fronthau Compression for Upink MMO Coud Radio Access etworks Yuhan Zhou, Yinfei Xu, Jun Chen, and Wei Yu Department of Eectrica and Computer Engineering, University of oronto, Canada

More information

ASummaryofGaussianProcesses Coryn A.L. Bailer-Jones

ASummaryofGaussianProcesses Coryn A.L. Bailer-Jones ASummaryofGaussianProcesses Coryn A.L. Baier-Jones Cavendish Laboratory University of Cambridge caj@mrao.cam.ac.uk Introduction A genera prediction probem can be posed as foows. We consider that the variabe

More information

CS229 Lecture notes. Andrew Ng

CS229 Lecture notes. Andrew Ng CS229 Lecture notes Andrew Ng Part IX The EM agorithm In the previous set of notes, we taked about the EM agorithm as appied to fitting a mixture of Gaussians. In this set of notes, we give a broader view

More information

Centralized Coded Caching of Correlated Contents

Centralized Coded Caching of Correlated Contents Centraized Coded Caching of Correated Contents Qianqian Yang and Deniz Gündüz Information Processing and Communications Lab Department of Eectrica and Eectronic Engineering Imperia Coege London arxiv:1711.03798v1

More information

Coalitions in Routing Games: A Worst-Case Perspective

Coalitions in Routing Games: A Worst-Case Perspective Coaitions in Routing Games: A Worst-Case Perspective Gideon Bocq and Arie Orda, Feow, IEEE arxiv:30.3487v3 [cs.ni] 27 Mar 206 Abstract We investigate a routing game that aows for the creation of coaitions,

More information

ESTIMATION OF SAMPLING TIME MISALIGNMENTS IN IFDMA UPLINK

ESTIMATION OF SAMPLING TIME MISALIGNMENTS IN IFDMA UPLINK ESTIMATION OF SAMPLING TIME MISALIGNMENTS IN IFDMA UPLINK Aexander Arkhipov, Michae Schne German Aerospace Center DLR) Institute of Communications and Navigation Oberpfaffenhofen, 8224 Wessing, Germany

More information

Coupling of LWR and phase transition models at boundary

Coupling of LWR and phase transition models at boundary Couping of LW and phase transition modes at boundary Mauro Garaveo Dipartimento di Matematica e Appicazioni, Università di Miano Bicocca, via. Cozzi 53, 20125 Miano Itay. Benedetto Piccoi Department of

More information

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries c 26 Noninear Phenomena in Compex Systems First-Order Corrections to Gutzwier s Trace Formua for Systems with Discrete Symmetries Hoger Cartarius, Jörg Main, and Günter Wunner Institut für Theoretische

More information

Sequential Decoding of Polar Codes with Arbitrary Binary Kernel

Sequential Decoding of Polar Codes with Arbitrary Binary Kernel Sequentia Decoding of Poar Codes with Arbitrary Binary Kerne Vera Miosavskaya, Peter Trifonov Saint-Petersburg State Poytechnic University Emai: veram,petert}@dcn.icc.spbstu.ru Abstract The probem of efficient

More information

8 Digifl'.11 Cth:uits and devices

8 Digifl'.11 Cth:uits and devices 8 Digif'. Cth:uits and devices 8. Introduction In anaog eectronics, votage is a continuous variabe. This is usefu because most physica quantities we encounter are continuous: sound eves, ight intensity,

More information

Research Article On the Lower Bound for the Number of Real Roots of a Random Algebraic Equation

Research Article On the Lower Bound for the Number of Real Roots of a Random Algebraic Equation Appied Mathematics and Stochastic Anaysis Voume 007, Artice ID 74191, 8 pages doi:10.1155/007/74191 Research Artice On the Lower Bound for the Number of Rea Roots of a Random Agebraic Equation Takashi

More information

Space-Division Approach for Multi-pair MIMO Two Way Relaying: A Principal-Angle Perspective

Space-Division Approach for Multi-pair MIMO Two Way Relaying: A Principal-Angle Perspective Gobecom 03 - Wireess Communications Symposium Space-Division Approach for Muti-pair MIMO Two Way Reaying: A Principa-Ange Perspective Haiyang Xin, Xiaojun Yuan, and Soung-Chang Liew Dept. of Information

More information

Expectation-Maximization for Estimating Parameters for a Mixture of Poissons

Expectation-Maximization for Estimating Parameters for a Mixture of Poissons Expectation-Maximization for Estimating Parameters for a Mixture of Poissons Brandon Maone Department of Computer Science University of Hesini February 18, 2014 Abstract This document derives, in excrutiating

More information

<C 2 2. λ 2 l. λ 1 l 1 < C 1

<C 2 2. λ 2 l. λ 1 l 1 < C 1 Teecommunication Network Contro and Management (EE E694) Prof. A. A. Lazar Notes for the ecture of 7/Feb/95 by Huayan Wang (this document was ast LaT E X-ed on May 9,995) Queueing Primer for Muticass Optima

More information

Precoding for the Sparsely Spread MC-CDMA Downlink with Discrete-Alphabet Inputs

Precoding for the Sparsely Spread MC-CDMA Downlink with Discrete-Alphabet Inputs IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL.*, NO.*, MONTH 2016 1 Precoding for the Sparsey Spread MC-CDMA Downin with Discrete-Aphabet Inputs Min Li, Member, IEEE, Chunshan Liu, Member, IEEE, and Stephen

More information

A GENERAL METHOD FOR EVALUATING OUTAGE PROBABILITIES USING PADÉ APPROXIMATIONS

A GENERAL METHOD FOR EVALUATING OUTAGE PROBABILITIES USING PADÉ APPROXIMATIONS A GENERAL METHOD FOR EVALUATING OUTAGE PROBABILITIES USING PADÉ APPROXIMATIONS Jack W. Stokes, Microsoft Corporation One Microsoft Way, Redmond, WA 9852, jstokes@microsoft.com James A. Ritcey, University

More information

Atomic Hierarchical Routing Games in Communication Networks

Atomic Hierarchical Routing Games in Communication Networks Atomic Hierarchica Routing Games in Communication etworks Vijay Kambe, Eitan Atman, Rachid E-Azouzi Vinod Sharma Dept. of Industria Engineering Management, IIT - Kharagpur, West Benga, India Maestro group,

More information

Coded Caching for Files with Distinct File Sizes

Coded Caching for Files with Distinct File Sizes Coded Caching for Fies with Distinct Fie Sizes Jinbei Zhang iaojun Lin Chih-Chun Wang inbing Wang Department of Eectronic Engineering Shanghai Jiao ong University China Schoo of Eectrica and Computer Engineering

More information

(f) is called a nearly holomorphic modular form of weight k + 2r as in [5].

(f) is called a nearly holomorphic modular form of weight k + 2r as in [5]. PRODUCTS OF NEARLY HOLOMORPHIC EIGENFORMS JEFFREY BEYERL, KEVIN JAMES, CATHERINE TRENTACOSTE, AND HUI XUE Abstract. We prove that the product of two neary hoomorphic Hece eigenforms is again a Hece eigenform

More information

Rate Adaptation Games in Wireless LANs: Nash Equilibrium and Price of Anarchy

Rate Adaptation Games in Wireless LANs: Nash Equilibrium and Price of Anarchy Rate Adaptation Games in Wireess LANs: Nash Equiibrium and Price of Anarchy Prasanna Chaporkar, Aexandre Proutiere, and Božidar Radunović Technica Report MSR-TR-2008-137 Microsoft Research Microsoft Corporation

More information

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law Gauss Law 1. Review on 1) Couomb s Law (charge and force) 2) Eectric Fied (fied and force) 2. Gauss s Law: connects charge and fied 3. Appications of Gauss s Law Couomb s Law and Eectric Fied Couomb s

More information

Theory of Generalized k-difference Operator and Its Application in Number Theory

Theory of Generalized k-difference Operator and Its Application in Number Theory Internationa Journa of Mathematica Anaysis Vo. 9, 2015, no. 19, 955-964 HIKARI Ltd, www.m-hiari.com http://dx.doi.org/10.12988/ijma.2015.5389 Theory of Generaized -Difference Operator and Its Appication

More information

sensors Beamforming Based Full-Duplex for Millimeter-Wave Communication Article

sensors Beamforming Based Full-Duplex for Millimeter-Wave Communication Article sensors Artice Beamforming Based Fu-Dupex for Miimeter-Wave Communication Xiao Liu 1,2,3, Zhenyu Xiao 1,2,3, *, Lin Bai 1,2,3, Jinho Choi 4, Pengfei Xia 5 and Xiang-Gen Xia 6 1 Schoo of Eectronic and Information

More information

Subspace Estimation and Decomposition for Hybrid Analog-Digital Millimetre-Wave MIMO systems

Subspace Estimation and Decomposition for Hybrid Analog-Digital Millimetre-Wave MIMO systems Subspace Estimation and Decomposition for Hybrid Anaog-Digita Miimetre-Wave MIMO systems Hadi Ghauch, Mats Bengtsson, Taejoon Kim, Mikae Skogund Schoo of Eectrica Engineering and the ACCESS Linnaeus Center,

More information

A SIMPLIFIED DESIGN OF MULTIDIMENSIONAL TRANSFER FUNCTION MODELS

A SIMPLIFIED DESIGN OF MULTIDIMENSIONAL TRANSFER FUNCTION MODELS A SIPLIFIED DESIGN OF ULTIDIENSIONAL TRANSFER FUNCTION ODELS Stefan Petrausch, Rudof Rabenstein utimedia Communications and Signa Procesg, University of Erangen-Nuremberg, Cauerstr. 7, 958 Erangen, GERANY

More information

Multiuser Power and Bandwidth Allocation in Ad Hoc Networks with Type-I HARQ under Rician Channel with Statistical CSI

Multiuser Power and Bandwidth Allocation in Ad Hoc Networks with Type-I HARQ under Rician Channel with Statistical CSI Mutiuser Power and Bandwidth Aocation in Ad Hoc Networks with Type-I HARQ under Rician Channe with Statistica CSI Xavier Leturc Thaes Communications and Security France xavier.eturc@thaesgroup.com Christophe

More information

Symbolic models for nonlinear control systems using approximate bisimulation

Symbolic models for nonlinear control systems using approximate bisimulation Symboic modes for noninear contro systems using approximate bisimuation Giordano Poa, Antoine Girard and Pauo Tabuada Abstract Contro systems are usuay modeed by differentia equations describing how physica

More information

Available online at ScienceDirect. IFAC PapersOnLine 50-1 (2017)

Available online at   ScienceDirect. IFAC PapersOnLine 50-1 (2017) Avaiabe onine at www.sciencedirect.com ScienceDirect IFAC PapersOnLine 50-1 (2017 3412 3417 Stabiization of discrete-time switched inear systems: Lyapunov-Metzer inequaities versus S-procedure characterizations

More information

FRIEZE GROUPS IN R 2

FRIEZE GROUPS IN R 2 FRIEZE GROUPS IN R 2 MAXWELL STOLARSKI Abstract. Focusing on the Eucidean pane under the Pythagorean Metric, our goa is to cassify the frieze groups, discrete subgroups of the set of isometries of the

More information

Stochastic Automata Networks (SAN) - Modelling. and Evaluation. Paulo Fernandes 1. Brigitte Plateau 2. May 29, 1997

Stochastic Automata Networks (SAN) - Modelling. and Evaluation. Paulo Fernandes 1. Brigitte Plateau 2. May 29, 1997 Stochastic utomata etworks (S) - Modeing and Evauation Pauo Fernandes rigitte Pateau 2 May 29, 997 Institut ationa Poytechnique de Grenobe { IPG Ecoe ationae Superieure d'informatique et de Mathematiques

More information

A New Algorithm for the Weighted Sum Rate Maximization in MIMO Interference Networks

A New Algorithm for the Weighted Sum Rate Maximization in MIMO Interference Networks A New Agorithm for the Weighted Sum Rate Maximization in MIMO Interference Networks Xing Li, Seungi You 2, Lijun Chen, An Liu 3, Youjian Eugene Liu Abstract We propose a new agorithm to sove the non-convex

More information

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 15, NO. 2, FEBRUARY

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 15, NO. 2, FEBRUARY IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 5, NO. 2, FEBRUARY 206 857 Optima Energy and Data Routing in Networks With Energy Cooperation Berk Gurakan, Student Member, IEEE, OmurOze,Member, IEEE,

More information

Rate-Distortion Theory of Finite Point Processes

Rate-Distortion Theory of Finite Point Processes Rate-Distortion Theory of Finite Point Processes Günther Koiander, Dominic Schuhmacher, and Franz Hawatsch, Feow, IEEE Abstract We study the compression of data in the case where the usefu information

More information

Equilibrium of Heterogeneous Congestion Control Protocols

Equilibrium of Heterogeneous Congestion Control Protocols Equiibrium of Heterogeneous Congestion Contro Protocos Ao Tang Jiantao Wang Steven H. Low EAS Division, Caifornia Institute of Technoogy Mung Chiang EE Department, Princeton University Abstract When heterogeneous

More information

A proposed nonparametric mixture density estimation using B-spline functions

A proposed nonparametric mixture density estimation using B-spline functions A proposed nonparametric mixture density estimation using B-spine functions Atizez Hadrich a,b, Mourad Zribi a, Afif Masmoudi b a Laboratoire d Informatique Signa et Image de a Côte d Opae (LISIC-EA 4491),

More information

SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS

SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS ISEE 1 SUPPLEMENTARY MATERIAL TO INNOVATED SCALABLE EFFICIENT ESTIMATION IN ULTRA-LARGE GAUSSIAN GRAPHICAL MODELS By Yingying Fan and Jinchi Lv University of Southern Caifornia This Suppementary Materia

More information

BP neural network-based sports performance prediction model applied research

BP neural network-based sports performance prediction model applied research Avaiabe onine www.jocpr.com Journa of Chemica and Pharmaceutica Research, 204, 6(7:93-936 Research Artice ISSN : 0975-7384 CODEN(USA : JCPRC5 BP neura networ-based sports performance prediction mode appied

More information

Coordination and Antenna Domain Formation in Cloud-RAN systems

Coordination and Antenna Domain Formation in Cloud-RAN systems Coordination and ntenna Domain Formation in Coud-RN systems Hadi Ghauch, Muhammad Mahboob Ur Rahman, Sahar Imtiaz, James Gross, Schoo of Eectrica Engineering and the CCESS Linnaeus Center, Roya Institute

More information

AALBORG UNIVERSITY. The distribution of communication cost for a mobile service scenario. Jesper Møller and Man Lung Yiu. R June 2009

AALBORG UNIVERSITY. The distribution of communication cost for a mobile service scenario. Jesper Møller and Man Lung Yiu. R June 2009 AALBORG UNIVERSITY The distribution of communication cost for a mobie service scenario by Jesper Møer and Man Lung Yiu R-29-11 June 29 Department of Mathematica Sciences Aaborg University Fredrik Bajers

More information

Cryptanalysis of PKP: A New Approach

Cryptanalysis of PKP: A New Approach Cryptanaysis of PKP: A New Approach Éiane Jaumes and Antoine Joux DCSSI 18, rue du Dr. Zamenhoff F-92131 Issy-es-Mx Cedex France eiane.jaumes@wanadoo.fr Antoine.Joux@ens.fr Abstract. Quite recenty, in

More information

Tracking Control of Multiple Mobile Robots

Tracking Control of Multiple Mobile Robots Proceedings of the 2001 IEEE Internationa Conference on Robotics & Automation Seou, Korea May 21-26, 2001 Tracking Contro of Mutipe Mobie Robots A Case Study of Inter-Robot Coision-Free Probem Jurachart

More information

Competitive Diffusion in Social Networks: Quality or Seeding?

Competitive Diffusion in Social Networks: Quality or Seeding? Competitive Diffusion in Socia Networks: Quaity or Seeding? Arastoo Fazei Amir Ajorou Ai Jadbabaie arxiv:1503.01220v1 [cs.gt] 4 Mar 2015 Abstract In this paper, we study a strategic mode of marketing and

More information

Oblivious Transfer over Wireless Channels

Oblivious Transfer over Wireless Channels 1 Obivious Transfer over Wireess Channes Jithin Ravi, Bikash Kumar Dey, Emanuee Viterbo arxiv:158664v1 [csit 4 Aug 15 Abstract We consider the probem of obivious transfer OT over OFDM and MIMO wireess

More information

Turbo Codes. Coding and Communication Laboratory. Dept. of Electrical Engineering, National Chung Hsing University

Turbo Codes. Coding and Communication Laboratory. Dept. of Electrical Engineering, National Chung Hsing University Turbo Codes Coding and Communication Laboratory Dept. of Eectrica Engineering, Nationa Chung Hsing University Turbo codes 1 Chapter 12: Turbo Codes 1. Introduction 2. Turbo code encoder 3. Design of intereaver

More information

Effect of transport ratio on source term in determination of surface emission coefficient

Effect of transport ratio on source term in determination of surface emission coefficient Internationa Journa of heoretica & Appied Sciences, (): 74-78(9) ISSN : 975-78 Effect of transport ratio on source term in determination of surface emission coefficient Sanjeev Kumar and Apna Mishra epartment

More information

Pricing Multiple Products with the Multinomial Logit and Nested Logit Models: Concavity and Implications

Pricing Multiple Products with the Multinomial Logit and Nested Logit Models: Concavity and Implications Pricing Mutipe Products with the Mutinomia Logit and Nested Logit Modes: Concavity and Impications Hongmin Li Woonghee Tim Huh WP Carey Schoo of Business Arizona State University Tempe Arizona 85287 USA

More information

A UNIVERSAL METRIC FOR THE CANONICAL BUNDLE OF A HOLOMORPHIC FAMILY OF PROJECTIVE ALGEBRAIC MANIFOLDS

A UNIVERSAL METRIC FOR THE CANONICAL BUNDLE OF A HOLOMORPHIC FAMILY OF PROJECTIVE ALGEBRAIC MANIFOLDS A UNIERSAL METRIC FOR THE CANONICAL BUNDLE OF A HOLOMORPHIC FAMILY OF PROJECTIE ALGEBRAIC MANIFOLDS DROR AROLIN Dedicated to M Saah Baouendi on the occasion of his 60th birthday 1 Introduction In his ceebrated

More information

THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES

THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES by Michae Neumann Department of Mathematics, University of Connecticut, Storrs, CT 06269 3009 and Ronad J. Stern Department of Mathematics, Concordia

More information

Combining reaction kinetics to the multi-phase Gibbs energy calculation

Combining reaction kinetics to the multi-phase Gibbs energy calculation 7 th European Symposium on Computer Aided Process Engineering ESCAPE7 V. Pesu and P.S. Agachi (Editors) 2007 Esevier B.V. A rights reserved. Combining reaction inetics to the muti-phase Gibbs energy cacuation

More information

XSAT of linear CNF formulas

XSAT of linear CNF formulas XSAT of inear CN formuas Bernd R. Schuh Dr. Bernd Schuh, D-50968 Kön, Germany; bernd.schuh@netcoogne.de eywords: compexity, XSAT, exact inear formua, -reguarity, -uniformity, NPcompeteness Abstract. Open

More information

Nearest Neighbor Decoding and Pilot-Aided Channel Estimation for Fading Channels

Nearest Neighbor Decoding and Pilot-Aided Channel Estimation for Fading Channels Nearest Neighbor Decoding and Piot-Aided Channe Estimation for Fading Channes A. Taufiq Asyhari, Tobias Koch and Abert Guién i Fàbregas arxiv:30.223v2 [cs.it] 8 Apr 204 Abstract We study the information

More information

Week 6 Lectures, Math 6451, Tanveer

Week 6 Lectures, Math 6451, Tanveer Fourier Series Week 6 Lectures, Math 645, Tanveer In the context of separation of variabe to find soutions of PDEs, we encountered or and in other cases f(x = f(x = a 0 + f(x = a 0 + b n sin nπx { a n

More information

Conditions for Saddle-Point Equilibria in Output-Feedback MPC with MHE

Conditions for Saddle-Point Equilibria in Output-Feedback MPC with MHE Conditions for Sadde-Point Equiibria in Output-Feedback MPC with MHE David A. Copp and João P. Hespanha Abstract A new method for soving output-feedback mode predictive contro (MPC) and moving horizon

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, we as various

More information

Cross-Layer Optimization of MIMO-Based Mesh Networks Under Orthogonal Channels

Cross-Layer Optimization of MIMO-Based Mesh Networks Under Orthogonal Channels Cross-Layer Optimization of MIMO-Based Mesh Networks Under Orthogona Channes Jia Liu, Tae Yoon Park, Y. Thomas Hou,YiShi, and Hanif D. Sherai Bradey Department of Eectrica and Computer Engineering Grado

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, as we as various

More information

HILBERT? What is HILBERT? Matlab Implementation of Adaptive 2D BEM. Dirk Praetorius. Features of HILBERT

HILBERT? What is HILBERT? Matlab Implementation of Adaptive 2D BEM. Dirk Praetorius. Features of HILBERT Söerhaus-Workshop 2009 October 16, 2009 What is HILBERT? HILBERT Matab Impementation of Adaptive 2D BEM joint work with M. Aurada, M. Ebner, S. Ferraz-Leite, P. Godenits, M. Karkuik, M. Mayr Hibert Is

More information

Radar/ESM Tracking of Constant Velocity Target : Comparison of Batch (MLE) and EKF Performance

Radar/ESM Tracking of Constant Velocity Target : Comparison of Batch (MLE) and EKF Performance adar/ racing of Constant Veocity arget : Comparison of Batch (LE) and EKF Performance I. Leibowicz homson-csf Deteis/IISA La cef de Saint-Pierre 1 Bd Jean ouin 7885 Eancourt Cede France Isabee.Leibowicz

More information

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract Stochastic Compement Anaysis of Muti-Server Threshod Queues with Hysteresis John C.S. Lui The Dept. of Computer Science & Engineering The Chinese University of Hong Kong Leana Goubchik Dept. of Computer

More information

An explicit Jordan Decomposition of Companion matrices

An explicit Jordan Decomposition of Companion matrices An expicit Jordan Decomposition of Companion matrices Fermín S V Bazán Departamento de Matemática CFM UFSC 88040-900 Forianópois SC E-mai: fermin@mtmufscbr S Gratton CERFACS 42 Av Gaspard Coriois 31057

More information

Sum Rate Maximization for Full Duplex Wireless-Powered Communication Networks

Sum Rate Maximization for Full Duplex Wireless-Powered Communication Networks 06 4th European Signa Processing Conference (EUSIPCO) Sum Rate Maximization for Fu Dupex Wireess-Powered Communication Networks Van-Dinh Nguyen, Hieu V. Nguyen, Gi-Mo Kang, Hyeon Min Kim, and Oh-Soon Shin

More information

Indirect Optimal Control of Dynamical Systems

Indirect Optimal Control of Dynamical Systems Computationa Mathematics and Mathematica Physics, Vo. 44, No. 3, 24, pp. 48 439. Transated from Zhurna Vychisite noi Matematiki i Matematicheskoi Fiziki, Vo. 44, No. 3, 24, pp. 444 466. Origina Russian

More information

Target Location Estimation in Wireless Sensor Networks Using Binary Data

Target Location Estimation in Wireless Sensor Networks Using Binary Data Target Location stimation in Wireess Sensor Networks Using Binary Data Ruixin Niu and Pramod K. Varshney Department of ectrica ngineering and Computer Science Link Ha Syracuse University Syracuse, NY 344

More information

Manipulation in Financial Markets and the Implications for Debt Financing

Manipulation in Financial Markets and the Implications for Debt Financing Manipuation in Financia Markets and the Impications for Debt Financing Leonid Spesivtsev This paper studies the situation when the firm is in financia distress and faces bankruptcy or debt restructuring.

More information

An Adaptive Opportunistic Routing Scheme for Wireless Ad-hoc Networks

An Adaptive Opportunistic Routing Scheme for Wireless Ad-hoc Networks An Adaptive Opportunistic Routing Scheme for Wireess Ad-hoc Networks A.A. Bhorkar, M. Naghshvar, T. Javidi, and B.D. Rao Department of Eectrica Engineering, University of Caifornia San Diego, CA, 9093

More information

ANALOG OF HEAT EQUATION FOR GAUSSIAN MEASURE OF A BALL IN HILBERT SPACE GYULA PAP

ANALOG OF HEAT EQUATION FOR GAUSSIAN MEASURE OF A BALL IN HILBERT SPACE GYULA PAP ANALOG OF HEAT EQUATION FOR GAUSSIAN MEASURE OF A BALL IN HILBERT SPACE GYULA PAP ABSTRACT. If µ is a Gaussian measure on a Hibert space with mean a and covariance operator T, and r is a} fixed positive

More information

Statistics for Applications. Chapter 7: Regression 1/43

Statistics for Applications. Chapter 7: Regression 1/43 Statistics for Appications Chapter 7: Regression 1/43 Heuristics of the inear regression (1) Consider a coud of i.i.d. random points (X i,y i ),i =1,...,n : 2/43 Heuristics of the inear regression (2)

More information

On the Optimality of Multiuser Zero-Forcing Precoding in MIMO Broadcast Channels

On the Optimality of Multiuser Zero-Forcing Precoding in MIMO Broadcast Channels On the Optimality of Multiuser Zero-Forcing Precoding in MIMO Broadcast Channels Saeed Kaviani and Witold A. Krzymień University of Alberta / TRLabs, Edmonton, Alberta, Canada T6G 2V4 E-mail: {saeed,wa}@ece.ualberta.ca

More information

Massive MIMO Communications

Massive MIMO Communications Massive MIMO Communications Trinh Van Chien and Emi Björnson Book Chapter N.B.: When citing this work, cite the origina artice. Part of: 5G Mobie Communications, Ed. Wei Xiang, an Zheng, Xuemin Sherman)

More information

(This is a sample cover image for this issue. The actual cover is not yet available at this time.)

(This is a sample cover image for this issue. The actual cover is not yet available at this time.) (This is a sampe cover image for this issue The actua cover is not yet avaiabe at this time) This artice appeared in a journa pubished by Esevier The attached copy is furnished to the author for interna

More information

A Novel Learning Method for Elman Neural Network Using Local Search

A Novel Learning Method for Elman Neural Network Using Local Search Neura Information Processing Letters and Reviews Vo. 11, No. 8, August 2007 LETTER A Nove Learning Method for Eman Neura Networ Using Loca Search Facuty of Engineering, Toyama University, Gofuu 3190 Toyama

More information

12.2. Maxima and Minima. Introduction. Prerequisites. Learning Outcomes

12.2. Maxima and Minima. Introduction. Prerequisites. Learning Outcomes Maima and Minima 1. Introduction In this Section we anayse curves in the oca neighbourhood of a stationary point and, from this anaysis, deduce necessary conditions satisfied by oca maima and oca minima.

More information

Uniformly Reweighted Belief Propagation: A Factor Graph Approach

Uniformly Reweighted Belief Propagation: A Factor Graph Approach Uniformy Reweighted Beief Propagation: A Factor Graph Approach Henk Wymeersch Chamers University of Technoogy Gothenburg, Sweden henkw@chamers.se Federico Penna Poitecnico di Torino Torino, Itay federico.penna@poito.it

More information