Research On Key Technologies In Multi-sink Wireless Sensor Networks

As a significant component of Internet of Things, Wireless Sensor Network(WSN) with its advanced information perception, data interaction and othercapabilities has attracted extensive attention of the researchers in recent years. WSNconsists of a large number of low-power, low-cost, and multi-functional microsensors which provide data acquisition and information processing service, and alsobe able to communicate with each other through the wireless channels. Hence, it canform a task-oriented network system on the basis of wireless communicationtechnology and self-organization method. However, densely deployed sensor nodesneed to transmit the data information to the user terminal or the control center throughthe sole sink in the traditional wireless sensor networks. This many-to-onetransmission mode put WSNs at enormous risks in many aspects such as energybalancing and reliability. Meanwhile, with the widespread applications of WSN, thesingle sink node structure can not meet the demand for rapid development of WSNs.Therefore, this paper explains how to solve the above problems by means ofincreasing the number of sink nodes, so that WSNs becomes more efficient, morestable, more robust, and easier to manage. This paper focusing on multi-sink nodeWSNs performs in-depth study from the routing protocol design, multiple sink nodesre-positioning technology and data aggregation strategies respectively, and alsoproposes relative algorithms. Achievements are summarized as follows:First, this paper proposes an optimal routing selection algorithm based on fuzzycomprehensive evaluation theory. By fuzzy comprehensive evaluating three factors,hops of sensor node to the sink node, the minimum residual energy of the path, aswell as the minimum average link quality, this algorithm provides a distributedoptimal multi-path routing corresponding to the sink node which is hierarchical,active and on-demand mixing and QoS in view. In consideration of humane thinking,routing in wireless sensor networks can provide a longer network lifetime, higherpacket delivery rate and less routing control information. Hence, in comparison with other multi-path routing protocols which don’t apply the fuzzy theory and takecomprehensive considerations to the effect factors, this algorithm will extend about2.8times of the lifetime of network and increase about14percent of the packetdelivery rate on average. But at the same time, it also reduces about37percent of thecontrol information during routing process.Secondly, this paper presents a relocation algorithm based on the centroidprinciple of multi-sink node. The algorithm utilizes the methods that make one-hopneighbors of the sink node as the system of particles and the number of packetstransmitted to the sink node as the particle quality, and regards the multiple sink nodeas the centroid of each system of particles by means of calculating the centroidposition in order to collaboratively approach and pinpoint the their correspondingoptimal position. In addition, the optimal position of the sink node can be adjustedaccording to the changes of network state, thus it can guarantee smaller averagenumber of hops and fewer average node energy. Therefore, compared with thealgorithms in which sinks are fixed, the relocation algorithm based on the centroidprinciple can extend about69percent of the network lifetime and improve about5percent of the packet delivery rate on average. It also has advantages over other sinkrelocation algorithms in the network lifetime by about18percent and enhances about6percent of the success rate of data transmission on average.Third, this paper proposes a data aggregation algorithm which is able to balancethe average energy consumption of sensor nodes and average data transmission delay.It first makes use of secondary fuzzy comprehensive evaluation to make relevantadjustment about the number of packets forward in a period through the originaloptimal routing to fit for data aggregation. As a result, new data forwarding paths canmore efficiently balance energy consumption of the sensor node and improve datatransmission reliability, as well as improving data aggregation performance byincreasing the overlap of the original paths. Moreover, the algorithm also can utilizethe states of the sensor nodes to dynamically adjust its current wait time ofaggregation, thus the data aggregation timing mechanism can effectively balance theaverage energy consumption of each sensor node, and reduce average data delay in network. Therefore, compared with algorithms that not involve data aggregation, itcan extend about one times of the network lifetime, but will reduce about3percent ofthe network packet delivery rate on average. It can also achieve better lifetime thanother data aggregation algorithms by about10percent, and improve about6percentof the packet delivery rate and about16percent of the aggregation rate as well asreducing about6percent of the average transmission delay.