Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 A Distributed Algorithm for Joint Sensing and Routing in Wireless Networks with Non-Steerable Directional Antennas Chun Zhang *, Jim Kurose +, Yong Liu.

Similar presentations


Presentation on theme: "1 A Distributed Algorithm for Joint Sensing and Routing in Wireless Networks with Non-Steerable Directional Antennas Chun Zhang *, Jim Kurose +, Yong Liu."— Presentation transcript:

1 1 A Distributed Algorithm for Joint Sensing and Routing in Wireless Networks with Non-Steerable Directional Antennas Chun Zhang *, Jim Kurose +, Yong Liu ~, Don Towsley +, Michael Zink + * IBM T.J. Watson Research Center + Dept of Computer Science, University of Massachusetts at Amherst ~ Dept of Electrical & Computer Engineering, Polytechnic University Nov 14, 2006 ICNP

2 2 Outline  motivation  problem formulation  distributed algorithm  result  summary

3 3 Multi-hop wireless sensor networks  sensor nodes  directional-antenna links  link capacity constraints 802.11 protocol: 2/5.5/11Mbps  energy constraints  energy supplied by solar panel sink A sink B applications: weather monitoring  performance metric  amount of information delivered to sinks

4 4 Interesting problem ? limited energy link capacities communication energy sensing energy sensing rate (information) radio layer application layer demand generator capacity generator more demand ? or more capacity? routing solution ? network layer

5 5 Our contribution  joint optimization problem formulation for energy allocation (between sensing, data transmission, and data reception), and routing  distributed algorithm to solve the joint optimization problem, with its convergence proved  simulation to demonstrate the energy balance achieved in a network of X-band radars, connected via point-to-point 802.11 links with non-steerable directional antennas

6 6 Related work  [Lin,Shroff@CDC04] [Eryilmaz,Srikant@ISC06]  joint rate control, resource allocation, and routing in wireless networks  our work further considers energy consumption for  data sensing  data reception

7 7 Outline  motivation  problem formulation  distributed algorithm  result  summary

8 8 Resource model  power resource  three power usages: data sensing, data transmitting, data reception  power is a convex and increasing function of data rate  constraint: consumption rate ≤ harvest rate  link capacity resource  constraint: link data rate ≤ link capacity  resource constraints satisfied by penalty functions

9 9 Goal : information maximization information modeled by utility function  : node i sensed and delivered data rate  node i collected information assumption: is a concave and increasing function

10 10 Optimization problem formulation s: sensing rates; X: data routes routes X deliver sensing rates s to data sink Joint sensing and routing problem

11 11 Transforming joint sensing/routing problem to routing problem with fixed demands i i’ wireless sensor network sensing link difference link sensing power -> reception power idea: treat data sensing as data reception through sensing link

12 12 Transformed problem fixed demand: maximum sensing rates; X: data routes routes X deliver maximum sensing rates to data sink Routing problem with fixed traffic demand

13 13 Outline  motivation  problem formulation  distributed algorithm  result  summary

14 14 Distributed algorithm: generalize [Gallager77] wired network algorithm  wired network  link-level resource constraint  wireless network  node-level resource constraint How to generalize from link-level to node-level?

15 15 Generalized distributed algorithm  generalize algorithm from wired network (link-level) to wireless network (node-level) repeat, until all traffic loaded on optimal path  each link locally compute gradient information  gradient information propagated from downstream to upstream in accumulative manner  routing fractions adjustment from non-optimal path to optimal path for generalized gradient-based algorithm: prove convergence provide step-size for routing fraction adjustment

16 16 Outline  motivation  problem formulation  distributed algorithm  result  summary

17 17 Simulation scenario From CASA student testbed  energy harvest rate: 7-13W  X-band radar-on power: 34W  radar-on rate 1.5Mbps  link-on trans power: 1.98W  link-on receive power: 1.39W  link-on goodput rate: as shown  Utility function 561234 111278910 171813141516 232419202122 293025262728 1Mb 2Mb 5.5Mb 2Mb 1Mb goodput rate

18 18 Optimization results for different energy harvest rates As power budget increases utility and sensing power increase communication power first increases, then decreases and flats out 561234 111278910 171813141516 232419202122 293025262728

19 19 Node level energy balance for different energy harvest rates power budget = 9W power budget = 13W  power rich network: max-min fair (single-sink) : sensing rates not affected by choice of utility functions  power constrained network: close to sink nodes spend less energy on sensing 561234 111278910 171813141516 232419202122 293025262728

20 20 Summary: a distributed algorithm for joint sensing and routing in wireless networks Goal : a distributed algorithm for joint sensing and routing Approach : 1.mapping joint problem to routing problem 2.proposed a distributed algorithm with convergence proof and step size Simulation to demonstrate energy balance for different energy harvest rates: 1.energy rich: proven max-min fairness (for single sink) 2.energy constrained: close-to-sink nodes spend more energy on communication, and thus less energy on sensing

21 21 Thanks ! Questions ?


Download ppt "1 A Distributed Algorithm for Joint Sensing and Routing in Wireless Networks with Non-Steerable Directional Antennas Chun Zhang *, Jim Kurose +, Yong Liu."

Similar presentations


Ads by Google