RESOURCE AND ENVIRONMENT AWARE SENSOR COMMUNICATIONS: FRAMEWORK, OPTIMIZATION, AND APPLICATIONS

摘要

Recent advances in low power integrated circuit devices,micro-electro-mechanical system (MEMS) technologies, andcommunications technologies have made possible the deployment oflow-cost, low power sensors that can be integrated to form wirelesssensor networks (WSN). These wireless sensor networks have vastimportant applications, i.e.: from battlefield surveillance systemto modern highway and industry monitoring system; from the emergencyrescue system to early forest fire detection and the verysophisticated earthquake early detection system. Having the broadrange of applications, the sensor network is becoming an integralpart of human lives. However, the success of sensor networksdeployment depends on the reliability of the network itself. Thereare many challenging problems to make the deployed network morereliable. These problems include but not limited to extendingnetwork lifetime, increasing each sensor node throughput, efficientcollection of information, enforcing nodes to collaborativelyaccomplish certain network tasks, etc. One important aspect indesigning the algorithm is that the algorithm should be completelydistributed and scalable. This aspect has posed a tremendouschallenge in designing optimal algorithm in sensor networks.This thesis addresses various challenging issues encountered inwireless sensor networks. The most important characteristic insensor networks is to prolong the network lifetime. However, due tothe stringent energy requirement, the network requires highly energyefficient resource allocation. This highly energy-efficient resourceallocation requires the application of an energy awareness system.In fact, we envision a broader resource and environment awareoptimization in the sensor networks. This framework reconfigures theparameters from different communication layers according to itsenvironment and resource. We first investigate the application ofonline reinforcement learning in solving the modulation and transmitpower selection. We analyze the effectiveness of the learningalgorithm by comparing the effective good throughput that issuccessfully delivered per unit energy as a metric. This metricshows how efficient the energy usage in sensor communication is. Inmany practical sensor scenarios, maximizing the energy efficient ina single sensor node may not be sufficient. Therefore, we continueto work on the routing problem to maximize the number of deliveredpacket before the network becomes useless. The useless network ischaracterized by the disintegrated remaining network. We design aclass of energy efficient routing algorithms that explicitly takesthe connectivity condition of the remaining network in to account.We also present the distributed asynchronous routing implementationbased on reinforcement learning algorithm. This work can be viewedas distributed connectivity-aware energy efficient routing. We thenexplore the advantages obtained by doing cooperative routing fornetwork lifetime maximization. We propose a power allocation in thecooperative routing called the maximum lifetime power allocation.The proposed allocation takes into account the residual energy inthe nodes when doing the cooperation. In fact, our criterion letsthe nodes with more energy to help more compared to the nodes withless energy. We continue to look at the problem of cooperationenforcement in ad-hoc network. We show that by combining therepeated game and self learning algorithm, a better cooperationpoint can be obtained. Finally, we demonstrate an example ofchannel-aware application for multimedia communication. In all casestudies, we employ optimization scheme that is equipped with theresource and environment awareness. We hope that the proposedresource and environment aware optimization framework will serve asthe first step towards the realization of intelligent sensorcommunications.