Optimal Power Allocation for Distributed BLUE Estimation with Linear Spatial Collaboration

Similar documents
Multiple Bits Distributed Moving Horizon State Estimation for Wireless Sensor Networks. Ji an Luo

Transmission Schemes for Lifetime Maximization in Wireless Sensor Networks: Uncorrelated Source Observations

QUANTIZATION FOR DISTRIBUTED ESTIMATION IN LARGE SCALE SENSOR NETWORKS

REDUCING POWER CONSUMPTION IN A SENSOR NETWORK BY INFORMATION FEEDBACK

Constellation Shaping for Communication Channels with Quantized Outputs

Censoring for Type-Based Multiple Access Scheme in Wireless Sensor Networks

SIGNAL STRENGTH LOCALIZATION BOUNDS IN AD HOC & SENSOR NETWORKS WHEN TRANSMIT POWERS ARE RANDOM. Neal Patwari and Alfred O.

SIGNAL STRENGTH LOCALIZATION BOUNDS IN AD HOC & SENSOR NETWORKS WHEN TRANSMIT POWERS ARE RANDOM. Neal Patwari and Alfred O.

certain class of distributions, any SFQ can be expressed as a set of thresholds on the sufficient statistic. For distributions

Rate-Constrained Distributed Estimation

Distributed Detection and Estimation in Wireless Sensor Networks: Resource Allocation, Fusion Rules, and Network Security

Towards control over fading channels

The Optimality of Beamforming: A Unified View

Optimal Quantization in Energy-Constrained Sensor Networks under Imperfect Transmission

The Effect of Memory Order on the Capacity of Finite-State Markov and Flat-Fading Channels

Diffusion based Projection Method for Distributed Source Localization in Wireless Sensor Networks

Capacity-achieving Feedback Scheme for Flat Fading Channels with Channel State Information

Constellation Shaping for Communication Channels with Quantized Outputs

Performance Analysis of Spread Spectrum CDMA systems

COOPERATIVE relay networks have recently attracted much

Computation of Bit-Error Rate of Coherent and Non-Coherent Detection M-Ary PSK With Gray Code in BFWA Systems

Morning Session Capacity-based Power Control. Department of Electrical and Computer Engineering University of Maryland

Diffusion LMS Algorithms for Sensor Networks over Non-ideal Inter-sensor Wireless Channels

Residual Versus Suppressed-Carrier Coherent Communications

Energy Efficient Estimation of Gaussian Sources Over Inhomogeneous Gaussian MAC Channels

Schur-convexity of the Symbol Error Rate in Correlated MIMO Systems with Precoding and Space-time Coding

On Design of Reduced-Order H Filters for Discrete-Time Systems from Incomplete Measurements

Lecture 7 MIMO Communica2ons

Estimation techniques

Binary Compressive Sensing via Analog. Fountain Coding

NOMA: An Information Theoretic Perspective

Vector spaces. DS-GA 1013 / MATH-GA 2824 Optimization-based Data Analysis.

5682 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 55, NO. 12, DECEMBER /$ IEEE

Optimal matching in wireless sensor networks

Lattice Reduction Aided Precoding for Multiuser MIMO using Seysen s Algorithm

Cooperative Communication with Feedback via Stochastic Approximation

Multi-User Gain Maximum Eigenmode Beamforming, and IDMA. Peng Wang and Li Ping City University of Hong Kong

EE 5407 Part II: Spatial Based Wireless Communications

Two-Way Training: Optimal Power Allocation for Pilot and Data Transmission

Dynamic Bandwidth Allocation for Target Tracking. Wireless Sensor Networks.

Optimum Power Allocation in Fading MIMO Multiple Access Channels with Partial CSI at the Transmitters

Fusion of Decisions Transmitted Over Fading Channels in Wireless Sensor Networks

ADAPTIVE CLUSTERING ALGORITHM FOR COOPERATIVE SPECTRUM SENSING IN MOBILE ENVIRONMENTS. Jesus Perez and Ignacio Santamaria

Applications of Robust Optimization in Signal Processing: Beamforming and Power Control Fall 2012

Comparisons of Performance of Various Transmission Schemes of MIMO System Operating under Rician Channel Conditions

On the Optimal Performance in Asymmetric Gaussian Wireless Sensor Networks With Fading

Improved Detected Data Processing for Decision-Directed Tracking of MIMO Channels

Optimal Data and Training Symbol Ratio for Communication over Uncertain Channels

Resource Allocation for Wireless Fading Relay Channels: Max-Min Solution 1 2

Distributed Binary Quantizers for Communication Constrained Large-scale Sensor Networks

2-D SENSOR POSITION PERTURBATION ANALYSIS: EQUIVALENCE TO AWGN ON ARRAY OUTPUTS. Volkan Cevher, James H. McClellan

COM Optimization for Communications 8. Semidefinite Programming

Lecture Notes 1: Vector spaces

Upper Bounds for the Average Error Probability of a Time-Hopping Wideband System

Per-Antenna Power Constrained MIMO Transceivers Optimized for BER

Tightened Upper Bounds on the ML Decoding Error Probability of Binary Linear Block Codes and Applications

Optimization of Multistatic Cloud Radar with Multiple-Access Wireless Backhaul

Binary Consensus Over Fading Channels: A Best Affine Estimation Approach

Vector Channel Capacity with Quantized Feedback

Rate and Power Allocation in Fading Multiple Access Channels

Diversity Combining Techniques

392D: Coding for the AWGN Channel Wednesday, January 24, 2007 Stanford, Winter 2007 Handout #6. Problem Set 2 Solutions

On Comparability of Multiple Antenna Channels

Multiuser Capacity in Block Fading Channel

Modulation & Coding for the Gaussian Channel

Convexity Properties of Detection Probability for Noncoherent Detection of a Modulated Sinusoidal Carrier

Input Optimization for Multi-Antenna Broadcast Channels with Per-Antenna Power Constraints

Energy Harvesting Multiple Access Channel with Peak Temperature Constraints

Asymptotic Distortion Performance of Source-Channel Diversity Schemes over Relay Channels

Chapter 4: Continuous channel and its capacity

Optimal Sensor Rules and Unified Fusion Rules for Multisensor Multi-hypothesis Network Decision Systems with Fading Channels

Data-aided and blind synchronization

Transmit Directions and Optimality of Beamforming in MIMO-MAC with Partial CSI at the Transmitters 1

Cooperative Spectrum Sensing for Cognitive Radios under Bandwidth Constraints

Nearest Neighbor Decoding in MIMO Block-Fading Channels With Imperfect CSIR

LIKELIHOOD RECEIVER FOR FH-MFSK MOBILE RADIO*

Sparse Sensing in Colocated MIMO Radar: A Matrix Completion Approach

Optimal Mean-Square Noise Benefits in Quantizer-Array Linear Estimation Ashok Patel and Bart Kosko

Random Access Sensor Networks: Field Reconstruction From Incomplete Data

On the Effectiveness of Multiple Antennas in. Distributed Detection over Fading MACs

Optimal Time Division Multiplexing Schemes for DOA Estimation of a Moving Target Using a Colocated MIMO Radar

Score Normalization in Multimodal Biometric Systems

The Effect of Channel State Information on Optimum Energy Allocation and Energy Efficiency of Cooperative Wireless Transmission Systems

Feedback Capacity of the First-Order Moving Average Gaussian Channel

Approximately achieving the feedback interference channel capacity with point-to-point codes

Minimum Feedback Rates for Multi-Carrier Transmission With Correlated Frequency Selective Fading

WIRELESS sensor networks (WSNs) comprise a large

Distributed Estimation via Random Access

Spatial and Temporal Power Allocation for MISO Systems with Delayed Feedback

Capacity of Memoryless Channels and Block-Fading Channels With Designable Cardinality-Constrained Channel State Feedback

Simulation studies of the standard and new algorithms show that a signicant improvement in tracking

DISTRIBUTED ESTIMATION IN SENSOR NETWORKS WITH IMPERFECT MODEL INFORMATION: AN ADAPTIVE LEARNING-BASED APPROACH

Optimal Transmit Strategies in MIMO Ricean Channels with MMSE Receiver

Problem Set 3 Due Oct, 5

A Systematic Description of Source Significance Information

The Impact of Correlated Blocking on Millimeter-Wave Personal Networks

ANALYSIS OF A PARTIAL DECORRELATOR IN A MULTI-CELL DS/CDMA SYSTEM

Using the Kalman Filter to Estimate the State of a Maneuvering Aircraft

Sigma-Delta modulation based distributed detection in wireless sensor networks

Expressions for the covariance matrix of covariance data

Transcription:

Optimal Power Allocation for Distributed BLUE Estimation with Linear Spatial Collaboration Mohammad Fanaei, Matthew C. Valenti Abbas Jamalipour, and Natalia A. Schmid Dept. of Computer Science and Electrical Engineering West Virginia University, Morgantown, WV, U.S.A. School of Electrical and Information Engineering University of Sydney, NSW, Australia IEEE ICASSP 04 Track : Waveforms and Signal Processing May 8, 04 M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Outline Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Outline Introduction and Problem Statement Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Introduction and Problem Statement What Is Distributed Estimation in WSNs? Spatially distributed sensors Observe their surrounding environment. Process their local observations. Send their processed data to a fusion center (FC). FC performs the ultimate global estimation. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 4/

Introduction and Problem Statement Background and Previous Works I n g v n w y g v r θ n n w w y r g v y r g v Fusion Center ˆ w y r Cui, Xiao, Goldsmith, Luo, and Poor, IEEE T-SP, 55 (9), September 007. Maşazade, Niu, Varshney, and eskinoz, IEEE T-SP, 58 (9), Sept. 00. n Ribeiro and Giannakis, IEEE T-SP, 54 (), March 006. w r Fanaei, Valenti, and Schmid, MILCOM, November 0. n y w g v M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 5/ g v

Introduction and Problem Statement Background and Previous Works I n g v n w y g v r θ n n w w y r g v y r g v Fusion Center ˆ w y r How can we etend this system model? Inter-sensor collaboration for signal processing. n Estimation of a vector of signals at the FC. w r n y w g v M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 5/ g v

Introduction and Problem Statement Background and Previous Works II n g v w y r n w g v θ n n w w w w w y r g v y r g v Fusion Center ˆ w y r S. ar and P. Varshney, Linear coherent estimation with spatial collaboration, IEEE Transactions on Information Theory, vol. 59, no. 6, pp. 5 55, June 0. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 6/

Introduction and Problem Statement Background and Previous Works II n g v θ n w θ n θ n w g v y r g v y r g v Fusion Center ˆ ˆ ˆ ˆ w y r θ w y r n I. Bahçeci and A. handani, Linear Linear Spatialestimation of correlated data in wireless sensor networks with Collaboration θ optimum power w allocation y and analogr modulation, IEEE Transactions on Communications, vol. 56, no. 7, pp. 46 56, July 008. n w θ w y r ˆ M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 6/ g g v v ˆ

.. θ Introduction and Problem Statement w y...... Our Proposed Research Problem n Linear Spatial Collaboration w θ n w θ w n θ n w w w w θ Goal of This Paper w. y y y g g g g v v v v r. r r y r r Fusion Center ˆ ˆ ˆ ˆ Adaptive power allocation to local sensors for estimation of the vector of signals observed by them at the FC of a WSN, when sensors perform spatial linear collaboration and communicate with the FC through parallel fading channels. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 7/

Introduction and Problem Statement Our Proposed Research Problem n Linear Spatial Collaboration w θ n w M θ w n θ n w w w M w θ Goal of This Paper w M... M y y. y M g g. g M v v v v M r r. r M Fusion Center ˆ ˆ ˆ ˆ Adaptive power allocation to local sensors for estimation of the vector of signals observed by them at the FC of a WSN, when sensors perform spatial linear collaboration and communicate with the FC through parallel fading channels. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 7/

Outline System Model Description Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 8/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Linear noisy observation of signals of interest θ [θ,θ,...,θ ] T with auto-correlation matri R θ E [ θθ T ]. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Sensors share their observations with each other through low cost, error-free links. Inter-sensor connectivity is modeled by an M-by- adjacency matri A. 0 0 0 A = 0 0 0 M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Sensors share their observations with each other through low cost, error-free links. Inter-sensor connectivity is modeled by an M-by- adjacency matri A. A is not necessarily symmetric. A j, j =. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Each connected sensor to the FC forms a linear combination of all local observations to which it has access: y j = w j,i i i= A j,i = Linear local processing (LINEAR COLLABORATION) as y = W Miing matri: W j,i = 0 if A j,i = 0 and W j,i = w j,i if A j,i =. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Average cumulative transmit power of the entire network is P Total = E [ y T y ] = E [ T W T W ] = Tr [ E [ yy T ]] = Tr [ W(R θ + R n )W T ] M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M Orthogonal channels corrupted by fading and (potentially correlated) AWGN. r = Gy + v = GW + v = GWθ + GWn + v G diag(g,g,...,g ) The FC has perfect knowledge of the channel fading coefficients. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description System Model Description n Linear Spatial Collaboration θ w n w M θ w n θ n w w M M... y y.... y M g g g M v r v r v M... r M Fusion Center ˆ ˆ ˆ ˆ θ w M The FC finds the best linear unbiased estimator (BLUE) for θ. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

System Model Description BLUE Estimation of θ at the FC r = Gy + v = GW + v = GWθ + GWn + v Since n v, given a realization of θ [θ,θ,...,θ ] T and G diag(g,g,...,g ): r {θ,g} N ( GWθ,GWR n W T G T ) + R v The BLUE estimator of θ at the FC is θ = (W T G T ( GWR n W T G T ) ) + R v GW W T G T ( GWR n W T G T + R v ) r The covariance matri of the BLUE estimator is R θ = (W T G T ( GWR n W T G T ) ) + R v GW S. M. ay, Fundamentals of Statistical Signal Processing: Estimation Theory, First Edition, NJ: Prentice Hall, 99 (Chapter 6). M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 0/

Outline Optimal Spatial-Collaboration Scheme Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Optimal Spatial-Collaboration Scheme Problem Formulation Problem Statement for Our Optimal Power-Allocation Scheme Derive the optimal miing matri W that minimizes the total distortion in the estimation of θ at the FC, given a constraint on the average cumulative transmit power of local sensors. minimize W Tr [W T G T ( GWR n W T G T ) ] + R v GW subject to Tr [ W(R θ + R n )W T ] P 0 M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Optimal Spatial-Collaboration Scheme Derivation of the Optimal Power Allocation I Lemma minimize Tr [W T G T ( GWR n W T G T ) ] + R v GW W subject to Tr [ W(R θ + R n )W T ] P 0 A lower bound on Tr [ R θ] can be found as Tr [ ] R θ [ ] Tr W T G T (GWR n W T G T + R v ) GW M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Optimal Spatial-Collaboration Scheme Derivation of the Optimal Power Allocation I Lemma minimize Tr [W T G T ( GWR n W T G T ) ] + R v GW W subject to Tr [ W(R θ + R n )W T ] P 0 A lower bound on Tr [ R θ] can be found as Tr [ ] R θ [ ] Tr W T G T (GWR n W T G T + R v ) GW maimize W Tr [W T G T ( GWR n W T G T ) ] + R v GW subject to Tr [ W(R θ + R n )W T ] P 0 M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Optimal Spatial-Collaboration Scheme Derivation of the Optimal Power Allocation II maimize Tr [W T G T ( GWR n W T G T ) ] + R v GW W subject to Tr [ W(R θ + R n )W T ] P 0 Lemma The above optimization problem is equivalent to minimize γ W,γ,Γ subject to Tr [ W(R θ + R n )W T ] P 0 ( Γ R n R n Tr[Γ] γ W T G T R v ) GW + R 0 where γ is a real scalar, and Γ is a symmetric -by- real matri. n M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 4/

Optimal Spatial-Collaboration Scheme Derivation of the Optimal Power Allocation III Lemma The above optimization problem is equivalent to minimize γ W,γ,Γ subject to Tr [ W(R θ + R n )W T ] P 0 ( Γ R n R n Tr[Γ] γ W T G T R v ) GW + R 0 where γ is a real scalar, and Γ is a symmetric -by- real matri. n IDEAS FOR PROOF: Woodbury matri inversion lemma. Schur s complement theorem. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 5/

Optimal Spatial-Collaboration Scheme Derivation of the Optimal Power Allocation III Lemma The above optimization problem is equivalent to minimize γ W,γ,Γ subject to Tr [ W(R θ + R n )W T ] P 0 ( Γ R n R n Tr[Γ] γ W T G T R v ) GW + R 0 where γ is a real scalar, and Γ is a symmetric -by- real matri. Linear programming with bi-linear matri-inequality constraints. Solved using numerical solvers such as PENBMI. n M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 5/

Outline Numerical Results Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 6/

Numerical Results Simulation Setup = 6 sensors are randomly and uniformly distributed in the twodimensional rectangle of [ 0, 0] [ 5, 5]. All sensors are connected to the FC, i.e., M =. σθ = : The variance of each component of θ. ρ i, j : Inter-sensor correlation coefficient. Monotonically decreases with the increase of the distance between sensors ( ) β di, ρ i, j e j β d i, j : Distance between sensors i and j. β = 6: Normalizing factor of the distances. β = : Controls the rate of the decay of the correlation coefficients. The covariance between the signals observed by sensors i and j R θi, j E[θ i θ j ] = σ θ ρ i, j M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 7/

Numerical Results Simulation Setup Homogeneous and equi-correlated vector of observation and channel noises with covariance matri defined as R n(v) = σ n(v) [( λn(v) ) I(M) + λ n(v) T ], σn(v) : Variance of each component of the vector of observation (channel) noises. λ n(v) : Constant correlation coefficients between each pair of distinct components of n (v). : Column vector of all ones with appropriate length. σ n = 0., σ v = 0.0, and λ n = λ v = 0.. The communication channels between local sensors and the FC have unit gain (g j = ). Each sensor collaborates with its q closest neighbors by sharing its local noisy observations through error-free, low cost links. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 8/

Numerical Results Effect of Linear Spatial Collaboration on Performance Total Estimation Distortion 0.9 0.8 0.7 5 0 Network -5 5 0 Network -5-0 -5 0 5 0 q=0 No Collaboration q= (Network ) q= (Network ) q= (Network ) q= (Network ) 0 4 6 8 0 4 Average Cumulative Transmission Power (P ) in db 0 Each sensor collaborates with its q closest neighbors by sharing its local noisy observations through error-free, low cost links. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 9/

Outline Conclusions Introduction and Problem Statement System Model Description Optimal Spatial-Collaboration Scheme 4 Numerical Results 5 Conclusions M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration 0/

Summary Conclusions We studied the effect of spatial linear collaboration on the performance of the BLUE estimator. The FC estimates the vector of spatially correlated signals observed by sensors. Even a small degree of connectivity and spatial collaboration could improve the quality of the estimators at the FC. The collaboration gain is more significant when the signals to be estimated have higher correlation. M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /

Conclusions Questions Thank You for Your Attention. Questions and/or Comments? M. Fanaei et al. Power Allocation for () Distributed BLUE Estimation with Linear Spatial Collaboration /