Research On Data Reliability Storage Based On Edge Layered Model And Robust Exponential Soliton Fountain Codes

Wireless sensor network is a a data-centric distributed group organization network,which is widely used in various types of environmental monitoring and disaster scenarios,such as volcanoes and earthquakes.Sensor nodes are deployed in the corresponding areas to collect and store relevant data,waiting for sink node to collect the data for further analysis and decision.However,in some bad scenarios,the sensor nodes in the network will become very fragile.The harsh living environment and limited energy will make the nodes fail,which seriously affects the collection of data integrity.Therefore,it's a very meaningful work to design an efficient,reliable data storage and collection strategy.At present,the research and application of network coding in wireless sensor networks has attracted much attention.In wireless sensor network,the durability of network data is effectively improved by introducing fountain codes technology.However,the traditional distributed coding strategy based on fountain codes may have "cliff effects" in the decoding process,which is due to the backlog of highly coded packets at the initial stage of network collection,and only a small amount of source data can be decoded if we can't collect enough data packets.Secondly,in a large-scale wireless sensor network,there are many nodes,and in order to ensure the overall security and reliability of the network,each node needs to perform high-complex coding and data forwarding work.However,the nodes in the network are very fragile and are easily damaged by the energy consumption and external environmental factors.Therefore,in some extreme disaster scenarios,there may even be a problem of node failure in a large area,which is seriously affects the persistent storage of data.In view of the problems of high data coding complexity,heavy node forwarding data burden,and low collection efficiency in the wireless sensor network mentioned above,this paper combines the charactersitics of the fountain codes,proposes a hierarchical coding framework of the edge layered model and an optimized degree distribution function to solve the problems.The methods can be described as follows:(1)Based on the similarity between Markov chain model and random walk,this paper proposes two different encoding schemes,one-way random walk and re-forward one-way random walk.At the same time,we use the characteristics of multicast delivery to further optimize the data encoding scheme to reduce the energy consumption of packet forwarding.These two encoding schemes use the characteristics of steady-state distribution to control the random walk step of encoded data packets in the network,and ensure that encoded data packets can be evenly dispersed in the network to complete the overall encoding of the network.(2)This paper improves the traditional decoding model,and proposes a hierarchical coding framework for the edge layered model.Sink node sends signals around the network to realize a simple hierarchical sequence of each nodes.In the collection phase,each node passes the data packets to the edge nodes of the network for storage according to the pre-divided hierarchical sequence.The sink node only needs to access the edge nodes to collect data,which greatly reduces the data collection cycle.Further,letting nodes send decoded data packets in order according to the hierarchical sequence,which further improves the efficiency of the data collection stage and effectively overcomes the "cliff effect" problem.(3)For a large-scale wireless sensor network scenario,we found that there will be some problems such as high encoding and decoding complexity,generating too many encoded data packets and so on if the data is still encoded according to the traditional robust soliton degree distribution.Therefore,in order to solve these problems,we want to optimize the robust soliton degree distribution.We combine the binary exponential degree distribution and the robust soliton degree distribution function,and then assign a certain coefficient to both degree distribution function to generate a new degree distribution function called exponentially robust soliton degree distribution function.Through simulation experiments,the optimized degree distribution function not only greatly reduces the overall coding complexity of the network,but also requires fewer coding packets to complete the encoding work without affecting the decoding rate,which is greatly improving the robustness of the network.