Outline Network structure and objectives Routing Routing protocol protocol System analysis Results Conclusion Slide 2

Transcription

1 2007 Radio and Wireless Symposium 9 11 January 2007, Long Beach, CA. Lifetime-Aware Hierarchical Wireless Sensor Network Architecture with Mobile Overlays Maryam Soltan, Morteza Maleki, and Massoud Pedram Department of Electrical Engineering University of Southern California Slide 1

2 Outline Network structure and objectives Routing protocol System analysis Results Conclusion Slide 2

3 Hierarchical Network Structure Sensor nodes Event Aggregation Relay ( (EAR EAR) )nodes Mobile Aerial Infrastructure overlay (MAIL)) nodes Base Station Flight Trajectory MAIL Node EAR Node Sensor Node M# of fmail nodes Base N # of EAR nodes Station Base Station Recurrent cycle Network Monitoring Lifetime (MoL) Flat architecture Slide 3

4 Objectives Analysis of concurrent controllable mobility and multi-hop routing in a multi-tier tier network Base Station Design and Analysis of Mobility-aware routing protocol Slide 4

5 Motivation Lower energy dissipation Longer lifetime. Energy consumption for wireless transmission: ε = ed β t d : Distance et : Energy dissipation for transmitting unit of data over unit of distance β : Path loss exponent Hierarchical network structure and multi-hop routing lowers energy dissipation. Mobility brings symmetry in battery depletion Slide 5

6 Bounded Hop Count Routing (BHR) Multi-hop routing between EAR nodes Less network delay and smaller storage size Shorter distance Less transmission power Dynamic Hop Count (DHC) vs. Initial Hop Count (IHC) mobility Routing delays transmission, propagation and queueing Bounded number of hops route if hop count <= H Storage, delay, and energy trade-off Slide 6

7 T 6 s 7 T 5 EAR Node Cluster and State Transitions in Each Cycle... Wait (S1) state Forward Chain γ S2 state μ 1... MAIL Coverage Area (Single Hop Routing) v... Backward Chain i j j+1 j+2 μ 2 k k+nc-1 k+nc k+nc+1 l... k + 1 k+nc-2 S3 state S4 state S5 S6 state S7 state (Nc -2) nodes T 7 s 1 T 1 Active States s7 s 6 s 2 s 5 s 3 T 2 s 5 3 T 4 s 4 T 3 Network of queues with vacation and variable service rate over time Slide 7

8 Queuing Analysis of Each EAR Node Wait (S1) state S2 state Forward Chain (Single Hop Routing) i j j+1 j+2 μ k 2... γ... v... Backward Chain k+nc-1 k+nc k+nc+1... μ 1 k + 1 k+nc-2 S3 state S4 state S6 state S7 state (a) Temporal variation of arrival rate for node i (b) Temporal variations of departure rate μ or μ 1 2 if Qi ( t ) > ( ) { t> T 0 1 Ui () t = For, Ai() t if Qi() t = 0 otherwise U ( t) = 0 i (c) () Temporal Variations of queue size l A i () t Ut i() Q i (t) + H ( H 1) (a) H 2 + μ 1 H.<μ 2 <μ 1 Δt B ( ) T1 T2 T3 T4 T5 T6 T7 μ H 1 2 μ ( H 1) 2 ( H 2 ) Δt B T1 T2 T3 T4 T 5 T6 7 L 1 μ 1 L 2 + μ μ L L 4 μ2 s 1 s 2 s 3 s 4 s 5 s 6 s 7 T 1 T2 T3 T 4 5 T T T T T Slide 8 (b) (c) M

9 Waiting Time in Queues EAR-to to-ear link delay: t link1 State durations: T i Avg. waiting in each EAR node: W t = W + ( 1/ μ ) link 1 1 t : link 1 Set as a variable T i Queueing analysis as a function of t link1 W as a function of t link1 System of two equations W Slide 9

10 Average Number of Hops for a Packet Transmission Hop count for packet delivery from an EAR node hi () t to a MAIL node Between 1 to H DHC seen by an EAR node during one cycle H 1 H Temporal average of DHC for a packet delivery 1 DHC t1 t2 t3 1 H 1 + h = H ( T1 + T2) + ( T4 + T5) + ( ) ( T3 + T6 ) T 2 M Δt s (H-1) Δt s (c+n ) Δ t tc c s (n ) Δt c s (H-1) Δts tn= Tcyc Slide 10

11 Network Delay Average delay for a packet to reach a MAIL node, D net D net ( h 1) 1 = h W + + μ μ 1 2 Waiting time in queues EAR-to-MAIL communication EAR-to-EAR communications Slide 11

12 Network Lifetime Average energy consumption for a packet delivery: E = ( h 1) ee + em e E : Avg. energy consumption for EAR-to-EAR communication e E M : Avg. energy consumption for EAR-to-MAIL communication Average # of packets generated during lifetime of the network: N T net 0 ( ) E N T N E T. 0 net E net = = E Slide 12

13 Max T vh,, Nc sys T = Min( T, c T ) sys net e Optimization Problem s.. t D D, B B, c C C net max p max M max Convex epigraph form: f 1 T f, f 1/ ct 1/ T sys e net Min f vhr,, c T e : MAIL endurance time T net : Network lifetime B p : Peak queue size C M M: : Recharge Cost c : # of charge occasions st.. ke f E 0, ρ v f c α E M 0 and D D, B B, c C C net max p max M max 3 ( T 3 1 ) e v Slide 13

14 Sample scenario: Simulation Results 1000 EAR nodes Q i (t) 1 Random distribution ib i Two MAIL nodes μ 2 e t for EAR-to-EAR: e t for EAR-to-MAIL: 10 (pjoul/bit/m 2 ) 1 T2 T3 : μ + μ1 μ2 s 1 s 2 s 3 s 4 s s 5 s 6 7 T T T T T Slide 14

15 Simulation Results Cont. Delay: Dnet 1 vm H 2 Lifetime: T net 1 H 2 Slide 15

16 Concluding Remarks Lifetime and delay aware deployment strategy A mobility-aware multi-hop routing protocol (Bounded hop count routing (BHR)) To control the trade off between delay, buffer size and lifetime Analysis of lifetime, delay, and buffer size Optimization problem formulation Packet level simulator Slide 16