Quality-of-information Oriented Resource Allocation In Wireless Sensor Networks

With the dramatic development of sensing,embedded computing and wireless communication techniques,wireless sensor networks(WSNs),which are small-size,low-cost and easy-deployment,have been applied to numerous areas,such as battlefield surveillance,environmental monitoring,animal tracking,medical equipment protection,security,business,and smart cities.Nevertheless,the success of these applications highly depends on the quality of information(QoI)provided by the WSNs.Furthermore,due to the limited available resources(e.g.,limited energy,frequency spectrum,data storage capacity,and battery capacity)and the dynamic characteristics of the network,low-quality information is prevalent in the applications of WSNs.The inaccurate and outdated information may degrade the cognitive level to the surrounding environment and further confuse the decision making,which will result in various undesirable results such as service interruptions,false alarms,and violation of privacy.Consequently,how to effectively utilize the limited available resources to improve the QoI is of great importance in WSNs.In this dissertation,several challenges to be addressed when developing the QoI management mechanisms in WSNs are illustrated at the beginning.Subsequently,the limitations of the existing QoI-oriented resource allocation algorithms in WSNs are outlined.Then,we devote to the design of resource allocation strategies to effectively utilize the network resources for the achievable QoI maximization.More specifically,the main contributions of this dissertation are summarized as follows:1)We investigate the energy-aware QoI maximization problem by jointly optimizing the sensor selection,sampling rate,packet-dropped rate,and transmit power in single-hop WSNs.By introducing the weight parameters,we first present a revenue-cost(RC)function which combines the optimization objectives of the QoI and the energy expenditure into a single objective to capture the trade-off between them.Then,a stochastic optimization programming is formulated to maximize the long-term average RC value subject to the network stability constraint.Exploiting the Lyapunov drift theory,we develop a low-complexity collaborative sensing and transmit power control(CSTPC)algorithm,which only requires the knowledge of instantaneous system state and can also guarantee the worst-case delay for each data packet.Additionally,we derive the bounds of time-averaged RC and queue length,and further quantify the RC-delay tradeoff as [O(1/V),O(V)] and can provide a flexible balance between them by simply adjusting V,where V is a system control parameter.Simulation results verify the effectiveness of our proposed algorithm and also demonstrate the proposed algorithm is feasible to improve the QoI while satisfying the energy dissipation requirements of applications under various environments and network conditions.2)We investigate the distortion minimization problem by jointly optimizing the sleep-wake scheduling and transmit power in energy harvesting single-hop WSNs.We consider two types of side information at the fusion center(FC): the noncausal information(i.e.,the energy harvesting profile and the channel power gains are known in advance up to a certain number of time slots)and the causal information(i.e.,the FC only has the knowledge of the instantaneous harvested energy and the instantaneous channel power gains and no knowledge of the future energy arrivals).In the case of attainable noncausal information and infinite battery capacity,a time-averaged mean-square-error(MSE)minimization problem over a finite horizon of multiple estimation periods is formulated,which is a mixed-integer nonlinear programming(MINLP).To circumvent the NP-hardness of this problem,we transform it into a tractable convex problem through canceling the nonlinear cross-multiplication terms and relaxing the integer constraints.Subsequently,a sleep-wake scheduling and power control algorithm(SSPCA)is proposed to obtain the optimal sleep-wake policy and power allocation for each sensor node(SN).We also consider the case where causal information is available and the capacity of battery storage is limited.The corresponding time-averaged MSE minimization problem over an infinite horizon is formulated.Exploiting the Lyapunov optimization theory combined with the idea of weight perturbation,we develop a suboptimal low-complexity sleep-wake scheduling and power control algorithm(SOSPA)to tackle the formulated problem.Simulation results are conducted to demonstrate the advantages of the proposed algorithms.3)We investigate the QoI maximization problem by jointly optimizing the data rate and transmit power in lifetime-constrained multi-hop WSNs.Firstly,the QoI at the sink node is characterized by virtue of the network utility,which quantifies the aggregated value of the data gathered from different SNs.Then,a network utility maximization(NUM)problem is formulated to maximize the QoI under the constraints of the network lifetime and the link capacity.To avoid oscillation among optimal solutions caused by the usage of multipath routing,the NUM problem is converted into an equivalent problem by exploiting the proximal optimization approach.Correspondingly,the transformed problem can be solved by the proposed proximal approximation based resource allocation algorithm(PARA)which has good features of fast convergence and low complexity.Moreover,we develop a successive convex approximation based algorithm(SCAA)to solve a nonlinear nonconvex D.C.(difference of convex functions)programming in the PARA.Finally,the convergence of the proposed algorithms is also proved theoretically.Simulation results verify the theoretical analysis and demonstrate the effectiveness of the proposed algorithms.