1 / 44

Using Movement Prediction To Reduce Energy Consumption in Wireless Communication

Using Movement Prediction To Reduce Energy Consumption in Wireless Communication. David K. Y. Yau Department of Computer Science Purdue University. Objective. Reducing energy consumption of battery powered devices, e.g., Laptops and Handhelds, in wireless networks.

chibale
Download Presentation

Using Movement Prediction To Reduce Energy Consumption in Wireless Communication

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Using Movement Prediction To Reduce Energy Consumption in Wireless Communication David K. Y. Yau Department of Computer Science Purdue University

  2. Objective • Reducing energy consumption of battery powered devices, e.g., Laptops and Handhelds, in wireless networks. • Wireless communication is power intensive. • Can we exploit node movement to reduce energy use in communication?

  3. Presentation Outline • Basic observation • Power saving strategy • System model • Heuristics • Simulation results • Conclusion • Future work

  4. Motivation • Wireless networks are getting popular. • Increasing interest in mobile ad hoc networks • Easy and low cost deployment • Mobility • No infrastructure • Highly dynamic • Problems • Routing – nodes keep moving in and out of the network. • Security – selfish, malicious, uncooperative nodes. • Scalability. • Limited battery life. • Network communication – major energy drainer. For handhelds over 50% of the battery life can be consumed by network interface card! • Improvements in battery technology - lifetime has increased. However, not to the extent to keep up with the increased energy requirement. • Needs software level energy saving strategies.

  5. Movement Prediction • Observation: Reduced distance between communicating peers ⇒ Reduced transmission power requirement ⇒ Energy saving. • Assuming network interface has transmission power control capability. • Single hop communication – obvious • Multi hop communication – expected

  6. Power Saving Strategy • If likely to move closer to the target, postpone communication for a future time. • Assuming application can tolerate some delay k. • Needs movement prediction • Based on movement history.

  7. Network Structure • Mobile nodes are moving within a rectangular plane. • We divide the network into virtual grids. • Each grid has a unique grid ID.

  8. Assumptions • Each node knows it's position – GPS. • Each mobile host maintains a sequence of n previous grid IDs. • Initial assumption – • target is fixed. • Every mobile node knows the target’s location. • Relax the fixed target assumption – • Both communicating peers are mobile.

  9. Mobility Model • Defines a stochastic process which tells us how a mobile node moves in a network. • Random waypoint mobility model • Wait for pause_time seconds • Pick a random new destination • Pick a random velocity • Move steadily to the chosen destination • Upon reaching the destination, repeat the steps 1 through 4 • Regular waypoint mobility model • Introduce regularity • Home – work – home model with occasional diversions • Choose new destination – not completely randomly • Two parameters – • Regularity r • Periodicity T

  10. Terminology • History of node h: Sh = {x1, x2, …, xn} • A window of size l (for i ≤ n-l+1): W(i,i+l-1) = {xi, xi+1, …, xi+l-1} • W(i,i+l-1) is a subsequence of Sh. • Distance between two grids i and j: d(i,j).

  11. Binary Distance (BD) Heuristic • Calculate the probability p that a mobile node will be in grid ID y within the next k time units as follows: Communicate immediately if p is less than some probability threshold pth. Else, postpone communication.

  12. A A t Problem With BD Heuristic Too coarse granular idea of distance – Counts only when the communicating node is in the same grid as the target.

  13. Binary Markov Distance (BMD) Heuristic • Based on order-m Markov model. • Calculate the probability that a mobile node will be in grid ID y within the next k time units as follows: • Problems: • Higher computational overhead. • Same coarse granularity problem as BD.

  14. Markov Distance (MD) Heuristic • Let R be the set of all possible routes that can be taken by the mobile node in the next k time units • Let R1, where R1 R, contain those routes in R that have at least one location closer to the target than the current distance. • Then, we calculate the probability that a mobile node will move closer to the target as: • If p ≥ pth, then we postpone the communication, else we communicate immediately. • Higher computational overhead • Distinguishes the distance between the node and the target on a finer level

  15. A t MD Heuristic - Example • Consider three possible paths of node A: • ρ1 moves closer to the target in the next two time steps. • ρ2 and ρ3 do not move closer to the target in the next two time steps. • If, these were the only options and A takes any of these paths with equal probability, then the probability that A will move closer to the target is: 1/3.

  16. Average Distance (AD) Heuristic • Calculate the average distance between a mobile node and the target over all windows of size k in the mobile node's movement history as: • If the current distance between the mobile node and the target is greater than avg, then the mobile node decides to postpone the communication, or else it communicates immediately. • Less Computational overhead • Takes into consideration the actual distance

  17. Analogy With Secretary Problem • Secretary problem: one must make anirrevocable choice from a number of applicants whose values are revealed only sequentially. • Our problem: we must choose one time step when a node communicates and once it communicates it is done. • Solutions to the secretary problem might help designing solutions to our problem.

  18. 37% Rule and The Least Distance (LD) Heuristic • Best-choice(r) Algorithm: reject the first r-1 candidates. Then accept the next candidate whose relative rank is 1 among the candidates seen till now. • Accepts the best candidate with probability 1/e ≈ 0.368. • Optimal solution. • Choose the time when the distance is the minimum seen till now. • LD Heuristic: find Minimum as: Postpone communication if current distance is greater that dmin, else communicate immediately.

  19. Single Threshold Solution • Select the first candidate whose valueexceeds a pre-specified threshold value. • Applicable only to the full information problem. • Parameters can be estimated from partial observation. • Average Distance heuristic – threshold is the average seen till now.

  20. One-bounce Rule • Keep checking values as long as theygo up. As soon as they go down we stop postponing any more and take thecurrent value. • postpone as long as the distance between the mobile host and the target is decreasing, and communicate as soon as the distance starts increasing. • Ignores the history other than the last value. • Use this idea along with AD heuristic.

  21. A A t Use of One Bounce Rule If a node is moving away from the target, average keeps decreasing at each time step and finally we choose the worst alternative. • Solution: Directional Average Distance Heuristic • Take direction of movement into consideration. • If at any point of time, moving away from the target, communicate immediately.

  22. Moving Target • Simple modifications to the heuristics proposed works for moving target. • Assume a mobile host s with location history Ss = {x1, x2, …, xn} wants to communicate with node r with location history Sr = {y1, y2, …, yn}. • MD heuristic: just define R and R1 with respect to Sr instead of y. • AD heuristic: Define average as: • LD heuristic: define minimum distance as:

  23. Preliminary Experiments • Number of Grids: 3 x 3 • Cost of single communication C(d) for distance d is d2. • 10000 repetitions. • Target Location: Randomly chosen for each run.

  24. Performance of BD Heuristic • Poor performance.

  25. Performance of BMD Heuristic

  26. Performance of MD Heuristic • With random waypoint mobility model.

  27. Performance of MD Heuristic • With regular waypoint mobility model.

  28. Performance of AD Heuristic

  29. Performance Comparison of BD, BMD, MD and AD.

  30. Simulation Experiments • Network size: 1500m x 1500m • Number of Grids: 3 x 3 • Number of nodes: 20 • Maximum speed: 10 m/s • Simulation time: 20000 seconds • Routing protocol: DSR • Propagation model: Two-ray ground. • Target Location: fixed at the center of the network.

  31. Performance of MD Heuristic for Varying Probability Threshold

  32. Performance of MD Heuristic for Varying Regularity

  33. Performance of AD Heuristic for Varying k

  34. Performance of AD and LD Heuristics for Varying Regularity

  35. Performance of LD Heuristic for Varying k

  36. Result for Mobile Target – Single hop Communication

  37. Result for Mobile Target – Multi hop Communication

  38. Observed Delay vs. Maximum Allowable Delay • We get higher energy saving by setting k higher, but without increasing the observed delay significantly.

  39. Energy Consumption Due to CPU Processing

  40. Comparison Among Heuristics

  41. Summary • Our strategy predicts a good time for communication, when some amount of delay is tolerable. • We postpone the communication until that point and then communicate. • Simulation results show significant energy saving.

  42. Conclusion • Wireless networking is rapidly emerging as the future communication technology • The components of an ad hoc network are mostly battery-powered handheld devices. • Limited battery life is an important issue in wireless networking. • We address the issue of exploiting node movement to reduce the energy cost of communication. • We can save more that 50% of the communication cost. • Computation cost of movement prediction algorithm is low.

  43. Future Work • Location information for moving target in ad hoc networks. • Considering transmission duration in predicting a good time for communication. • Optimal way to divide the network into grids.

  44. Thank you!

More Related