Research On Lifetime Optimization Algorithms For Wireless Sensor Networks

Wireless sensor networks were originated in the military field and mainly used in military surveillance system. Now, low-cost wireless sensor networks have been used in environmental and meteorological monitoring, flood warning, farm management, intelligent home, intelligent transportation and other fields. Therefore, it gets more and more attention from academia and industry.In wireless sensor networks, network lifetime is one of the most important indicators to judge the network performance good or bad. It is a major research. But wireless sensor nodes are mostly battery-powered and have limited power. Once one node exhausts energy and is disabled, it may affect network data routing, even break up the network to shorten network lifetime. Therefore, to avoid excessive energy consumption and premature failure of some nodes, the algorithms of wireless sensor networks must take power-saving into account to prolong the network lifetime. Therefore, it is necessary to research on the lifetime optimization problem of wireless sensor networks.This thesis focuses on network lifetime optimization problem. It uses some methods such as shortest path methods in graph theory, optimization methods and power control methods to solve lifetime optimization problem of wireless sensor networks under different scenarios and prolong the network lifetime. The main work and achievements of the thesis are as follows:1. Lifetime optimization problem of static wireless sensor networks is researched and lifetime optimized routing algorithm based on shortest path tree (LORA_SPT) is proposed. The weight function with energy for transmitting data between network nodes, residual energy of own nodes and residual energy of neighbor nodes are established. The nodes are divided into standard nodes and warning nodes. The link weights with different types of nodes use different weighting factors. Finally, dijkstra algorithm is used to construct the shortest path tree. All nodes transmit data along the shortest path tree to sink node.2. The scheme of optimal lifetime in static wireless sensor networks is researched and lifetime maximization distributed algorithm based on Newton method (LMDA_NM) is proposed. The algorithm establishes and decomposes the network optimization model, introduces nonnegative slack variables and logarithmic barrier functions, and establishes node optimization model with local information. Newton method is used to solve the model and obtain optimal values of network maximum lifetime and link transmission data amount.3. The problem of sink node mobility is researched and lifetime optimized distributed algorithm for mobile sink node (LODA_MSN) is proposed. The algorithm considers the mobility of sink node as discrete movement. Then the optimization problem is divided into lifetime maximization problems of several networks when sink node is static. Each lifetime maximization problem of static wireless sensor network is solved by LMDA_NM algorithm and finally the optimal value of network maximum lifetime is obtained when sink node moves several times.4. With nearest-neighbor algorithm and power control algorithm, the preferable transmission power in node uniform distribution network and cluster optimal transmission power in non-uniform distribution network are researched. Nearest-neighbor power control algorithm for optimizing lifetime in single cluster (NPCAOL_SC) and nearest-neighbor power control algorithm for optimizing network lifetime in multi-clusters (NPCAOL_MC) are proposed. For uniformly distributed wireless sensor networks, sink node saves the entire network topology information, uses various nearest-neighbor distances algorithm to measure node density, and determines preferable communication distance. Then preferable transmission power is calculated with Friss free space model. Finally sink node broadcasts to inform nodes that they transmit data with the transmission power. If nodes can not find neighbor node with preferable transmission power, they use maximum transmission power. For non-uniformly distributed wireless sensor networks, k-means algorithm is used to determine number of clusters and the corresponding network nodes in each cluster. In the same cluster, nodes use preferable transmission power of cluster to communication. Nodes between different clusters use maximum transmission power to communication.5. When the node can not measure the distance to its neighbor nodes, the transmission power change scheme with residual energy is researched, distributed power control algorithm for optimizing lifetime based on shortest path tree (DPCAOL_SPT) is proposed. Considering energy for transmitting data and neighbor nodes’residual energy, the new weight function and three power attenuation models such as stepwise attenuation model. γn order attenuation model and linear attenuation model are introduced. Finally distributed asynchronous Bellmam-Ford algorithm is used to construct the shortest path tree. All nodes transmit data along the shortest path tree to sink node.6. When node transmission power is fixed, lifetime optimization scheme under transmission power constraint is researched and distributed power control algorithm for optimizing lifetime based on subgradient algorithm (DPCAOL_SA) is proposed. The algorithm analyzes the conditions such as node energy constraint when transmission power is fixed. It establishes the maximum network lifetime model. To solve the model, distributed transmission power iteration and subgradient algorithm are used. Nodes obtain the minimum transmission power set needed to communicate with neighbors, randomly select the current transmission power from the set. receive the parameter information of neighbor nodes, and distributed compute node maximum lifetime with subgradient algorithm. After several iteration calculations, DPCAOL_SA can obtain the local optimal value of network maximum lifetime, local optimal transmission power of each node and current data forwarding probability.A number of illustrative simulations are given to show the effectiveness of proposed algorithms. Finally, the conclusion and future work are presented.