| Wireless Sensor Network (WSN) is an integrated intelligent information system with function of information collection, information transmission and information processing. WSNs are composed of a large number of sensor nodes with very limited computation capability, power and memory. It is difficult to replace or charge the sensor nodes due to the working conditions and their own structural limitations. Therefore, it is significant to do the research about energy efficient routing protocol.Firstly, the characteristics, development status, structure and heterogeneity of WSNs are summarized in this paper, then several typical routing protocols are analyzed and the performance of these protocols are compared. Finally, an energy efficient clustering routing protocol is proposed for the heterogeneous networks, in which the heterogeneity of energy and computation capability of sensor nodes is considered. In this protocol, the BS (Base Station) runs the cluster selection algorithm, once the cluster heads and associated clusters are found, the BS broadcasts a message that contains the cluster head ID for each node. Since the BS is not limited in computing capability, communication capability and energy, it can reduce the energy consumption of sensor nodes during the processing of clustering. Besides, in order to reduce the communication energy consumption of sensor nodes caused by reporting their residual energy to the BS frequently, we suggest that the BS can estimate the residual energy of each sensor node by the model of energy consumption.Based on the new routing protocol, we focus on the key technology of cluster routing protocol-clustering algorithm and we propose two new clustering algorithms which are complementary in their performance. In the CSBK (the cluster head selection algorithm based on the improved K-means for heterogeneous WSNs), we take into account the residual energy, geographical location, energy consumption rate and data processing capability of sensor node to evaluate whether it can be selected as candidate cluster head or not. This algorithm is simple and converges fast. In the CSBAP (the cluster head selection algorithm based on the Affinity Propagation for heterogeneous WSNs), a communication cost function is designed as the input of AP to get a high precise cluster structure, which takes into account the residual energy, geographical location and data processing capability of sensor node.The simulation results show that both of the two new algorithms can prolong the network lifetime effectively compared with the congener algorithm named BSK-means. But CSBAP is more likely to be used in the scene of applying long network lifetime and the CPU processing capability of the BS is fast, while CSBK is more likely to be used in the scene of applying the high-level network coverage and the CPU processing capability of the BS is slow. |
