Research Of Key Algorithms For Sensor Networks

Sensor network, which is made by the convergence of sensor, micro-electro-mechanism system and network technologies, is a novel technology about acquiring and processing information and can be used to transfer informations to the terminal users. It demonstrates comprehensive application prospect in many fields, such as military,medicine,space exploration ,industry,civilian and home networks.Scientists predict that sensor network will become an important part of daily life in our new era.Being limited by the application environments,sensor networks which use wireless communicstion tecnology are different from traditional networks because the battery being used can not be recharged.Therefore, making full use of energy efficiently and prolonging life time becomes the main issue of sensor network designing.But,present protocals and theoeries of wireless,ad_hoc and bluetooth networks can not be used to sensor work directly so that many people focus on the research of models,protocols and theories being used to sensor networks.The main content of this paper can be divided into four parts.Firstly, based on LEACH(Low Energy Adaptive Clustering Hierarchy),the restrictive condition of a optimal cluster number formula and a new optimal cluster number formula is proposed and it is approved that the life time of the sensor network is longest under this condition.According our simulation results that the energy dissipation is associate with sensor nodes distribution structure in the condition of same cluster number.The evener that sensor nodes ditribute,the less that whole energy dissipates.At the same time ,an uniform clustering algorithm of the sensor network is provided and its effectiveness is verified by simulation.Seconly,a model of relay sensor network being used to communication among moving objects is shown and communication efficiency is anlynized.Further more,an adjacent boundary condition about the performance comparision of push and pull algorithm based on mail box is obtained,which gives the theoretical reference forselecting one of the two algorithms.To improve the communication efficiency further,a region partion scheme is investigated and a formula related to it is given.Thirdly,packets schedule and routing algorithms are researched on the bases of bluetooth theoretical background and it is proved that AFP algorithm is better than RR algorithm. On the other hand,the new algorithms of packets schedule and routing are designed and the simulation results show that they have higher efficiency,less energy dissipation and longer life time than trodational algorithms.Fourthly, the question about Time Synchronization is discussed. Time Synchronization is necessary to any distributed systems and it is very important when sensor network needs to confirm the time sequences of events taking place.For example,estamating the speeds of moving objects by detecting vibration time sequences , ascertaining the position by detecting the acoustic speeds produced by moving objects,avoiding to store redundant data gathered by sensor nodes at same time to the same event.At present, the question of Time Synchronization faces on many challenges including resouces scarcity,time error resulted by data packets routing delay and clock excursion due to using cheap crystal.The clock being made of cheap crystal is easy to excursion and excursional rate is unknown.In this paper ,an algorithm of time synchronization driven by event is proposed on the bases of low energy and the error of the algorithm is analyzed.According probability,a theoretical method of estimating the error resulted by transfer delay is put forward .The algorithm mentioned above is suitable for the requirements of time synchronization and saves energy remarkably.Finally,considering different features of sound wave and structures of underwater marine exploration networks,we study the problem of the optimal cluster number calculation and give an approach of calculating the optimal cluster number.On the other hand,we describe a condition to determine the redundant sensor nodes.