Topology Control Algorithms For Energy-efficient In Multi-level Wireless Sensor Networks

With the development of scientific technology, wireless sensor networks (WSNs) havebeen a hot topic in the world. Meanwhile, topology control algorithms, which are used toconstruct an optimal or sub-optimal topology, are key technologies for the WSNs to collectand transfer messages efficiently.In this dissertation, three efficient topology control algorithms are proposed based on thehomogeneous model, heterogeneous model and lossy links model respectively, which is toreduce energy consumption of nodes and extend the network lifetime.1. A tree-based topology construction algorithm with probability distribution andcompetition in the same layer (PCLT) is proposed based on the homogeneous WSNs to lowerthe communication packets and energy consumption. A theorem of probability distributionabout the number of nodes in each layer is discussed through theoretical verification. Besides,weighted values of nodes are calculated by broadcasting messages in the PCLT, and with acompetition method in the same layer, the best parent nodes are selected. Furthermore, astrategy of secondary waken is given when the node needs are to be waked up by itsprobability distribution.2. A clustering-tree algorithm based on the energy forecast (CTEF) is proposedcombining the network practicality for the load balancing issue based on the heterogeneousWSNs. The actual average residual energy of the r-th is estimated through the ideal averageenergy and an energy difference in CTEF. Meanwhile, cluster heads are selected by thecommunication cost, and the non-cluster head nodes are conducted to join cluster through theenergy, link reliability and packet loss rate. At last, some transferring nodes are chosen bydividing the cluster area through cross axis composing clustering tree to decrease the load ofcluster heads.3. In the environment of lossy links, a link prediction topology control algorithm basedon the Markov chain (LPA) is presented. Above all, a node with the highest energy is chosenin each small area, and the link quality in the next round is predicted through a link-statetransition matrix. The matrix is composed by values of received signal strength indicator and signal noise rate. Simultaneously, some nodes are selected as active nodes by fuzzymathematics. At last, the shortest path method is used to combine the small area and the sinkfor constructing a spanning tree.Simulation results have proved the efficiency of the researches, which is proposed aboveand compared with the existing algorithms; they can save more energy and prolong thenetwork lifetime.