Loading...

Distributed Cross-layer Optimization in Wireless Sensor Networks

Mirghasemi, Hamed | 2009

618 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 39815 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Pakravan, Mohammad Reza
  7. Abstract:
  8. In this thesis, our aim was to assess the resource allocation problem in wireless sensor netowroks and to optimize network utility and lifetime. To achieve these aims, it is nesseccary to solve resource allocation problem through optimization methods. Since the central methods need a central node to manage resource allocation, they can not be implemented in sensor networks because the nodes neighbors to this central node drain energy very quickly. Also, since central methods require transmission of all information about network topology to this central node, every change in network topology result in transmission of new network topology informations and therefore, overload for frequent chaging networks can be huge. These two remarks demonstrate the importance of distributed solution for resource allocation problem. One of the distributed methods to solve optimization problem is dual decomposition which its performance is proved for optimization problems. In this thesis, the resource allocation of a CDMA-based wireless sensor network to minimize the total power transmission is modeled through two cases: layer-layer and cross-layer optimization problem. These optimization problems are decomposed horizontally (i.e. cross the nodes) and vertically (i.e. cross the layers) through dual decomposition method. Since the capacity of each link is a function of the power transmission of the other links, in each step of our obtained algorithm, message passing from a node to all other nodes is required and this message passing can result in a huge overload. To make our algorithms more practical, some methods for overload reduction are proposed. The better method for energy consumption management in wireless sensor networks is to maximize the lifetime of the network. The resource allocation problem to maximize the utility and liftetime of the network is a complicated optimization problem and to decompose this problem, I used slack variables (i.e. local versions of lifetime variable) which make our problem decoupled across the nodes. As a result of solving this problem, I could maximize the network utility and lifetime and also, each node is aware of the network lifetime and therefore, it knows when the network topology changes and it should restart the algorithm. One of the practical issues which considered in this thesis is packet loss probability which could affect the optimum route, transmission rate and lifetime. In addition to utility and lifetime, another parameter optimizied in this thesis is the total link delays. Therefore, our algorithm not only will be optimum for utility and lifetime but also for delay.
  9. Keywords:
  10. Resources Allocation ; Cross Layer Optimization ; Network Lifetime ; Wireless Sensor Network ; Dual Decomposition

 Digital Object List

 Bookmark

No TOC