Research On Method Of Residual Energy Real-time Monitoring In Wireless Sensor Networks

Wireless sensor network (WSN) is energy-constrained. Residual energy real-time monitoring (RERM) is a fundamental way for getting knowledge of the life-time of WSN. RERM has great meanings in designing and validating the energy-efficient algorithms/protocols. However, current studies on RERM cannot reflect the real situation of WSN. Therefore, we did the RERM study in a completely application-oriented way. We found that the energy consumption of a node has direct relationship with its true throughput. Therefore, we discuss the energy consumption problem from the change of true throughput.Our contributions of this thesis include the followings:(1) Both the application-oriented general network model (AGNM) and application-oriented electromagnetic environment model (AEEM) for RERM are proposed. Experimental result and related analysis show that the AGNM model can effectively characterize the topologies, running-time mechanism and running states of personal area network (PAN), and the AEEM model can effectively characterize the dynamic change of electromagnetic environment. Therefore, we chose these two models as the research basis for the application-oriented RERM.(2) A scalable framework for residual energy real-time monitoring (SFRERM) is proposed. Experimental result and related analysis show that this framework is suitable for being deployed practically; also, comparing with other frameworks/algorithms, the SFRERM framework has great advantages in many aspects (such as scalability, robusticity, precision, time-delay, energy-efficiency and simplicity etc.). Therefore, it can provide fundamental infrastructure support for the application-oriented RERM researches.(3) An energy consumption self-sense model based on characteristic analysis of communication and process is proposed. With the support of SFRERM, this model effectively characterizes the real situation of energy consumption in WSN. Moreover, the costs of self-sense are lower than that of other methods. However, the original information used by this model must be precise, complete and punctual.(4) An energy consumption self-inference model based on Fuzzy Bayesian Network (FBN) is proposed. Especially, this model elaborately characterizes the probability causality among retransmission and related factors. With the support of SFRERM, this model can effectively infer the energy consumption of WSN by utilizing the imprecise information. For the retransmission phenomenon is fully taken into account, the precision of result is higher than that of other models and methods. However, the original information used by this model must be complete and punctual.(5) An energy consumption self-prediction model based on Fuzzy Dynamic Bayesian Network (FDBN) is proposed. With the support of SFRERM, this model can effectively predict the energy consumption of continuous time slices by utilizing the imprecise and incomplete information. For a mechanism of utilizing history information (delayed information etc.) is introduced into this model, the precision of prediction is improved greatly. In summary, this model elaborately processes the imprecise, incomplete and delayed original information. Then, a favorable precision of prediction can be achieved.