Research On Data Gathering Technology For Low-Duty-Cycle WSNs Based On Compressive Network Coding

With the progress of sensor technology,micro-electromechanical system,wireless communication technology and the decrement of the cost of manufacturing in recent years,wireless sensor networks(WSNs)have enjoyed a speedy development.It have been widely deployed in the fields such as military,industrial control,environmental monitoring.transportation,health care and warehouse management.WSNs are an information acquisition system,which are commonly composed of a large number of cheap micro-sensor nodes deployed in the monitoring area.These nodes form a multi-hop self-organizing network via wireless communication configuration,to cooperatively perceive,acquire,handle and transmit information of various sensing targets in the network coverage area.Energy efficiency is the primary design goal of WSNs,nodes are usually designed to operate on low-duty-cycle mode to reduce the energy wasted by idle listening.However,low-duty-cycle operation will bring about data transmission throughput,latency,reliability and other challenges.Compressive network coding based schemes can reduce the amount of data transmission and balance the node load,thus suitable for the nodes' duty-cycle operation.This dissertation combines the compressive network coding technology to study the data collection for low-duty-cycle WSNs from the network topology,efficiency and transmission reliability aspects.(1)The dissertation designed a new data collection scheme for clustered duty-cycle WSNs and proposed a data gathering scheme based on compressive network coding for clustering low-duty-cycle WSNs.First,according to the requirements of greenhouse environment monitoring such as dense nodes,wide coverage,we proposed a practical fixed-cluster duty-cycle WSNs considering the characteristics of greenhouse equipment control.In order to reduce the network energy consumption,a data collection mechanism based on compressive network coding is proposed for the fixed cluster duty-cycle WSNs.By analyzing the effect of duty-cycle mode,packets' length,number of nodes in the cluster and the coding coefficiency,we form a three-stage mechanism:network packets re-encoding,cluster heads measurements and sink node decoding reconstruction.Experimental results show that this mechanism can fully consider node duty mode,balance network load,prolong network life cycle,and realize efficient data collection task for duty cycle WSNs.(2)The dissertation proposed an adaptive data transmission schemes for WSNs based on compressive network coding.In each data collection epoch,each measurement result is formed by linear encoding process in the measurement-formation path and the sink node can evaluate the reconstruction quality of the received measurement results.Therefore,the proposed scheme resolves the contradiction in existing research that the sparsity of network monitoring data is constant and the value of the sparsity is known before starting each data gathering epoch.Furthermore,an adaptive termination rule is proposed for sparse measurement methods,as well as the proof of its validity.Considering the low-duty-cycle operation of nodes,we established an optimal mathematical model to describe the sparse-measurement formation,and it has several advantages:does not require time synchronization between nodes,improved the efficiency of measurement formation and reduced the number of control messages in the network.Since the mathematical model can be reduced to an optimal Steiner tree problem and belongs to the NP-complete problem,we also proposed a greedy iterative algorithm to solve this problem.Simulation experiments and actual network deployments show that the proposed adaptive data aggregation method is effective,and the energy consumption,the number of control messages is lower than the existing schemes.(3)The dissertation proposed a multipath routing scheme based on compressive network coding.This scheme can improve the reliability of data transmission and synchronously reduce energy consumption within the network.And,comparing to the multipath routing scheme solely based on network coding,it retains the advantages,but also avoid the"all-or-nothing" problem.Another typical feature of this scheme is that nodes' encoding and decoding operation work over the finite field,thus solving the data cumulative problem which is caused by operation over the real field in previous studies.This dissertation also proved the equivalence of the proposed compressed sensing measurement matrix and random linear network coding coefficient matrix.Simulation and test-bed experiment show the performance advantages of proposed scheme in data transmission reliability and energy efficiency.