Research For Distributed Resource Allocation And Queue Scheduling In Wireless Sensor Networks

Wireless sensor networks are widely used in many fields, such as military,industry, medical treatment, transportation and so on. As a product of the times, it willoccupy an increasingly important position in the future. Such networks typically havea large number of low-power and low-cost sensor nodes. The source nodes areresponsible for collecting the data in the environment, and passing it to the relaynodes, by which the data will be passed to the sink nodes via multi-hops. However,the nodes in the sensor networks mainly gain energy through battery with limitedpower, and limited network channel capacity limits the transmission rate anddistribution of data in the network. Therefore, it is critical to ensure quality of servicefor wireless sensor networks by allocating resource efficiently and queue scheduling,so that the network has a stable performance. In addition, with the rapid growth ofdata traffic and the scale of current communication networks, it has been quitedifficult for the station node to process the data from all sink nodes in a centralizedway. Therefore, to solve the network problem for resource allocation and queuescheduling in a distributed way not only can greatly improve network performance,but also greatly enhance the speed of the network to solve practical problems. So it isa very significant research topic to solve network resource allocation and queuescheduling problem with limited resource in a distributed way.Based on the full understanding of the present WSNs research status ondistributed resource allocation, queue scheduling and distributed algorithm at homeand abroad, we in-depth study flow control optimization problems and distributedalgorithms with the constraints of limited resource and stable queze in wireless sensornetworks. The main work and innovation are as follows:(1) We develop a flow control problem with the constraints of node energy andlink capacity in the wireless sensor networks with fixed single-path routing. Withoutconsidering the actual characteristics that the WSN mainly powered by the battery with limited energy, the traditional network utility maximization problem onlyconsided the constraint about link capacity. This article will consider the impact of thelimited resources for overall performance in the wireless sensor network. In addition,the traditional method to solve this problem is dual-subgradient algorithm that has adisadvantage of slow convergence and sensitive step-size selections, so we developeda distributed Newton method with second-ordet convergence to improve theperformance of algorithm. The study show that this problem can be solved bytransforming and matrix splitting, and the proposed algorithm can achieve quadraticconvergence rate while requiring almost the same communication overhead as thesubgradient-based dual-decomposition algorithm. Moreover, it is shown bysimulations that the proposed algorithm is much better than the subgradient-baseddual-decomposition algorithm in terms of both runtime and the total number ofiterations.(2) Based on the research studies point (1), we develop a joint multi-hop routingand flow control utility maximization problem to achieve utility proportional fairness.Different from the traditional problem, this problem not only include the constraintsof node energy and link capacity, but also consider the best route to select.Furthermore, we choose "pseudo utility function" as the objective function to achieveproportional fairness. In particular，it should be noted that, the optimization problemscenario is completely different from the previous network problems (such asmultipath, full row rank for the coefficient matrix, block diagonal for Hessian matrix,etc.), resulting in a completely different algorithm and execution from the previousnetwork. Through a series of questions equivalence transformation, reorganization ofthe variables, using of matrix splitting, we show this problem can still be solvedefficiently. Finally, the simulations demonstrated the performance of distributedNewton algorithm.(3) We proposed a utility optimal scheduling problem with constraints of limitedresource, queue stability and no-underflow in wireless sensor networks to ensurenetwork permance. In addition, in order to study the impact of link interference on theperformance of the wireless sensor network, this paper further introduce the concept of content matrix and extend the model into a queue scheduling for WSNs with link’sinterference. To solve the two issues efficiently, we designed a Perturbed Max-weightalgorithm base on Lyapunov optimization methods and duality theory. Finally, weprove the permance of this algorithm theoretically and verify the convergence byusing the Matlab software. The simulation show that the algorithm can not onlyensure the stability of the network, but also obtain a solution with a error within (1/V) and a queue backlog that satisfies (V).