Research On Reliable Data Delivery In Wireless Sensor Networks

In recent years, with the in-depth research of related technologies and the improvement in hardware, there are more and more deployed wireless sensor networks in practice and show its potential. As one important problem in wireless sensor networks, reliable data transmission determines how data packets can be delivered to sink successfully through multi-hop routing. However, the reliability of data transmission in wireless sensor network is bad due to the negative effect of environment, signal interference and energy exhaustion on wireless communication.Redundant data transmission is the main technique to guarantee reliable data transmission, such as multipath or retransmission, the main idea of which is to gain reliability by adding redundancy to data transmission. Since most applications that need high reliability are also delay sensitive, we use multipath as a basis to achieve reliable data transmission in this thesis. Introducing much redundant transmission, multipath is highly restricted by the inherent characteristics of wireless sensor network, such as energy-critical, large-scale, dynamic and bandwidth limited. These characteristics have brought many challenging problems in research and application of reliable data transmission.In this thesis, we deeply study some key problems of path construction and data transmission based on multipath technique. These problems are highly related with each other and have direct impact on the effeteness and energy efficiency of multipath. However, multipath still cannot fully take effect due to the limitation of current research on the communication energy consumption model, scalability, coding format, traffic and bandwidth allocation. Aiming at the inherent characteristics of wireless sensor network and the limitation of current work, this thesis studies these key problems comprehensively.Aiming at the energy inefficiency and coupling problem of multiple paths, we first study the problem of constructing reliability constrained multiple disjoint paths with minimum energy consumption and prove it is a NP-hard problem, then give the near-optimal solution on how to construct paths based on the analysis of accurate communication energy consumption model. In order to reduce path interference, we further take advantage of the multi-frequency characteristic of current radio and propose an energy optimizing and interference avoidance multipath construction algorithm EIMM-FA, which construct and maintain multiple paths according to the available frequencies and node density. Simulation results show that EIMM-FA reduces path interference considerably and the desired reliability can be achieved with minimum energy consumption.In order to improve scalability problem of multipath construction, we make use of the strong correlation of geographically proximate sensors and propose a new reliable clustering protocol CREED. In order to balance energy consumption among intra-cluster nodes, cluster size of CREED is determined based on the multi-rate capabilities of 802.11 a/b/g technologies. In cluster head selection, a dynamic cluster head backup scheme EDDS is presented to be ready to switch to new cluster head whenever the old one is unavailable, which guarantee the robustness of cluster head with less energy consumption. In inter-cluster routing, CREED includes a multipath routing protocol to guarantee reliable data delivery from cluster head to sink. This protocol allocates traffic to multiple paths according to nodes' residual energy and path hops to balance energy consumption among different paths. Simulation results show that CREED improve the scalability of multipath construction and optimizes intra-cluster and inter-cluster energy consumption, providing an energy-efficient and robust cluster topology support for reliable data transmission.To make good use of multipath and achieve reliable transmission effectively, we need further study the control on multipath data transmission. Traditional multipath routing transmits multiple original copies of each packet concurrently along multiple paths, which may increase the dependency on individual packet. Aiming at this problem, we combine multipath routing with a promising network coding technique in this thesis. This scheme includes a group method of data packet on source node. By coding per-group packets into multiple independent new packets and transmitting along multiple paths, intermediate node recoding all received packets belonging to the same group, sink only need to receive part of these packets to recover the original data. This scheme can reduce the bad effect caused by channel instability and improve the reliability of data transmission. In this scheme, the number of path needed by each group is a key parameter which exhibits the amount of energy needed by multipath rouing through combining with network coding. We propose an approximate method to compute this parameter based on the de Moivre-Laplace central limit theorem. Analysis and simulations show that network coding can guarantee the same reliability while reduce redundancy as much as possible, and fit the dynamic and large-scale characteristic well in wireless sensor networks.In multipath data transmission, we further study the corresponding congestion control problem. Improper traffic allocation will cause and deteriorate congestion, which result in lossing effect of multipath. Aiming at this problem, we propose a multipath based congestion control framework COTA+CODEM to control congestion effectively while keeping the constraint of desired reliability. In order to avoid congestion, COTA predict the potential congested area according to the use frequency and residual energy of path nodes, based on which it allocates traffic to multiple paths to avoid traversing the potential congestion region. Once congestion happens inevitably, a runtime traffic adjustment CODEM is used to detect congestion and divert traffic from the congested area to other paths. Based on accurate congestion metric and rules, CODEM can mitigate congestion timely while achieving the desired reliability. Simulation results. verify the distinguished performance of COTA+CODEM in reliability, delay and network throughput.In congestion, we still need to consider the fairness control on data transmission. Aiming at the incompleteness of current fairness definition and unpracticalness of current fairness control strategy, we extend the traditional fairness definition based on the redundantly deployment characteristic of wireless sensor network, and propose a practical fairness control model CFRC. In CFRC, a low-cost credit computation algorithm is proposed for each source node to compute its credit locally based on the sensed area of itself and its neighbors. Credit is the basis for bandwidth allocation. Furthermore, an interferer aware fair rate allocation algorithm is proposed in CFRC to fully achieve fairness. Simulation results show that CFRC can allocate bandwidth based on the credit of data sources, and achieve fairness in accordance with the characteristic of wireless sensor network.In summary, our work present solutions to several key problems of the multipath construction and its scalability, coding of multipath data transmission, congestion and fairness control in reliable data transmission, and has academic and practical value for advancing the theory and practicability of reliable data transmission in wireless sensor network.