Network Lifetime Cross-Layer Optimization In Wireless Sensor Networks

Energy is critical for typical wireless sensor networks (WSN) and how to minimize en-ergy consumption and maximize network lifetime (NL) are big challenges for WSN. Whilemost of these research efforts have been spent on solving the NL maximization (NLM) prob-lem in a numerical manner, there are few research efforts on analyzing what this solution isand how the NL depends on optimization variables, which provide an innovative insight intothe NL. To find out, we utilize Karush-Kuhn-Tucker (KKT) optimality conditions to solvethe NLM problem analytically. In this paper, we investigate a cross-layer design approachfor minimizing energy consumption and maximizing network lifetime (NL) of a multiple-sources and single-sink (MSSS) WSN with energy constraints. The optimization problemfor MSSS WSN can be formulated as a mixed integer convex optimization problem with theadoption of time division multiple access (TDMA) in medium access control (MAC) layer,and it becomes a convex problem by relaxing the integer constraint on time slots. Impactsof data rate, link access and routing are jointly taken into account in the optimization prob-lem formulation. This dissertation focuses on solving this convex optimization problem byanalytical approaches. The main contributions are as follows:1. The existing theoretical method available for studying cross-layer NLM is extended byproviding analytical duality techniques, i.e., utilizing the KKT optimality conditions.The KKT conditions for NLM problem are derived.2. In linear single-source and single-sink (SSSS) WSN, the globally optimal solutionsare derived by solving the KKT conditions.3. In linear multiple-source and single-sink (MSSS) WSN, the suboptimal solutions arederived by solving the KKT conditions under fixed routing scheme and D&C ap-proach.4. In linear MSSS WSN, from the definition of NL and KKT conditions, we obtain thetight suboptimal solutions that enable all the nodes to exhaust their energies at the same time. In addition, the globally optimal solutions can also be derived by solvingthe KKT conditions directly while the source rates are very low.5. In planar SSSS WSN, from the definition of NL and KKT conditions, we derive thetight suboptimal solutions that enable all the nodes to exhaust their energies at thesame time.6. In planar MSSS WSN, an iterative algorithm based on a decomposition and combina-tion (D&C) approach is proposed to obtain a suboptimal NL. In the D&C approach,the optimization problem for MSSS network is decomposed into multiple optimizationproblems for SSSS networks. The SSSS optimization problems are optimally solvedindividually and combined to produce a suboptimal solution for MSSS network.