Research On Network Topology Control Algorithms In Software-defined Wireless Sensor Networks

Based on the solution of software-defined network(SDN),a software-defined wireless sensor network(SD-WSN),which is a new wireless sensor network(WSN),decouples the controlling function from the forwarding function,facilitating the flexibility of network management.The network energy efficiency(EE)is still a key factor to restrict WSN applications in SDWSN,in which Topology Control(TC)plays an important role in the improvement of network EE,and it is one of paramount issues in WSN.In the SDWSN for Internet of Things(IoT)applications,the intensive deployment of sensor nodes and the diversity of services increase the dynamics of the network topology,which challenges the real-time issues of the traditional TC technologies.This dissertation is devoted to research dynamic TC problems of SDWSN for the real-time challenges,focus-ing on the control algorithms for variable and controllable scenarios of the network topology,which mainly includes: the impact analysis and modeling of relay selection(RS)in SDWSN,time-varying network TC algorithms based on the link EE in SDWSN,variable power control algorithm based on the energy-efficient routing protocol in SDWSN,and time-varying topology-oriented dynamic routing algorithm in SDWSN.The studied algorithms enable the network topology to be reconfigured when encountering abnormal events or network state changes by making full use of controller-centralized controlling and software programming.The main achievements and contributions of this dissertation are listed as follows:1)For the problem of the impact of RS in the path between the source node and the destination node on the network lifetime,an energy-efficient topology model based on the minimum-energy path is established,which includes the compound best RS criteria for different scenarios and the energy-efficient relay search region(RSR)model.The former includes two patterns based on the minimum distance and the minimum-impact probability,which applied for an obstacle-less scenario and an obstacle scenario,respectively.The latter proposes the energy-efficient RSR concept,and extends the definition of the optimal relay position(ORP)by analyzing the energy-efficient RS characteristics.Meanwhile,an optimal hop distance(OHD)is further defined for the energy-efficient RSR model,upon which the impact of OHD on the network lifetime is analyzed and discussed.To validate the effectiveness of the established model,an energy-efficient RS algorithm based on the established model is proposed.Simulation results demonstrate that the proposed algorithm has the advantages of low complexity and easy implementation,and can effectively extend the network lifetime as the network size increases.2)For the problem of the impact of network-state changes in SDWSN on the dynamic topology mainte-nance,two time-varying TC algorithms,referred to as the link-EE-based common sensor node(CSN)cluster network TC algorithm and the link-EE-based software-defined sensor network(SSN)relay network TC algo-rithm,are proposed.The former algorithm,which is applied for the single-hop cluster network composed of CSNs in the SDWSN,reduces the interference between the cluster members by using the cluster heads to allo-cate the related resources for its members,and improves the average network EE by using the CSN-gateway node to select the cluster head with the high link EE dynamically.The latter algorithm,which is applied for the multi-hop relay network composed of SSNs in the SDWSN,constructs the original network topology by using the energy-efficient RSR model and the Dijkstra's shortest path model(DSPM),respectively,and models the network-state transition process as a Markov decision process(MDP),upon which the best RS policies can be achieved by the value-iteration learning method in order to maximally reduce the impact of the shared channel interference.Simulation results show that both two proposed time-varying TC algorithms can effectively improve the corresponding network EE when the SDWSN topology changes.The CSN-cluster network TC algorithm can increase the network EE by reducing the cluster size,which is verified in the SD-WSN platform implementation,and the SSN-relay network TC algorithm can achieve a better network EE by the best RS policy in the energy-efficient RSR model than that in the DSPM model.3)For the problem of the variable power allocation for any data-source node during the dynamic topol-ogy maintenance of SDWSN,a variable power control algorithm based on the energy-efficient routing pro-tocol is proposed.The controller chooses an energy-efficient link,and dynamically allocates the transmit power with adjustment of the transmission time for a source node according to its energy load and the neigh-boring link energy loads while guaranteeing the quality of service(QoS).The proposed control algorithm is implemented by the energy-efficient RS sub-algorithm and the centralized relay scheduling sub-algorithm.The former sub-algorithm utilizes the energy-efficient RSR model to construct an original network topology,upon which the valid link set for a node can be determined,as well the related link priorities can be calculated.The latter sub-algorithm is used to allocate the link rate meeting the link QoS requirements in the valid link set,and dynamically select the relay node and adjust the transmit power and the transmission time for the source node.Simulation results show that during the dynamic topology maintenance of SDWSN,the pro-posed control algorithm can reduce the average network energy consumption by timely adjusting the transmit power and the transmission time for the source node,and further extend the network lifetime by balancing the link energy loads.4)For the problem of the impact of different energy-efficient RS policies in the routing protocols on the network performances under the time-varying topology scenario,an energy-efficient RS based dynamic routing algorithm is proposed,in which it designs a dynamic link weight considering both the link reward and the link cost in terms of the energy-efficient RSR model.The proposed routing algorithm analyzes and compares the network performances obtained by the different energy-efficient RS policies through adjusting the proportional coefficient about the link reward and the link cost.The designed link reward is dynamic and related to the node residual energy and allocated resources,while the designed link cost is relatively static and related to the relative locations of nodes in the energy-efficient RSR model.Simulation results show that increasing the proportional coefficient of the link reward in the designed link weight can improve the network EE.As compared with the routing algorithm based on the DSPM model,the proposed algorithm has a greater flexibility,which can be suitable for IoT applications with the time-varying topology.