High-efficiency Data Transmission Methods In Wireless Sensor Networks

Technological advances in the development of sensing, embedded computingand low-power wireless communication technologies have made possible scenariosin which a large number of battery-powered sensor nodes are deployed in amonitoring field to collaboratively perceive, collect and process the sensinginformation by wireless communication way, A wireless sensor network (WSN) iscomprised of many sensor nodes, which are connected by a self-organized form andsend the processed information to the base station by a multi-hop form. WSNs arewidely applied in many fields, such as environmental monitoring, medical care,intelligent transportation and military defense. One distinguishing feature of WSNsis data-centred. Hence, how to efficiently and effectively transmit data is animportant research problem in WSNs. A naive approach is to instruct all nodes inthe sensor network to send their readings at regular intervals to a base station. Thisapproach, however, imposes a serious burden on storing and transmitting data and issignificantly inconsistent with one of the major concerns for WSNs: energyconservation. Because sensor nodes have limited battery lifetime, and radiotransmission is the primary consumer of energy, excessive transmission quicklydepletes batteries, rendering the nodes useless. Therefore, how to reduce the amountof data while not compromising the mission of applications is a fundamentalresearch problem. Facing this challenge, this dissertation considers dataactrloagmnsopmrirteihssmssiisoo. nnT hapenro mdb leaicmno cn ogonefts rtiibowunit reioclenossns ot rfos tleh, nsor networks from two aspects, datais adnidss erptraotpioons easr e dasif feorlelonwt s:strategies andFirst, this dissertation proposes a disconnected piecewise linear compressionalgorithm GDPLA. Power consumption is a critical problem affecting the lifetimeof wireless sensor networks. To reduce the cost of storage, transmission andprocessing of time series data generated by sensor nodes, the need for morecompact representations of time series data is compelling. Although a large numberof data compression algorithms have been proposed to reduce data volume, theiroffline characteristic or super-linear time complexity prevents them from beingapplied directly on time series data generated by sensor nodes. Motivated by these observations, this dissertation propose an optimal online algorithm GDPLA (GreedyDisconnected Piecewise Linear Approximation) for constructing a disconnectedpiecewise linear approximation of a time series which guarantees that the verticaldistance between each real data point and the corresponding fit line is less than orequal to ε. GDPLA not only generates the minimum number of segments toapproximate a time series with precision guarantee, but also only requires lineartime O(n) bounded by a constant coefficient6. The low cost characteristic of ourmethod makes it a proper choice for resource-constrained WSNs. Extensiveexperiments on two real datasets have been conducted to demonstrate the superiorcompression performance of our algorithm.Second, A disconnected piecewise curved compression algorithm3D-FSS(Three Dimensional-Feasible Solution Space) is proposed in this dissertation. Tothe best of our knowledge, this algorithm is the first one to use disconnectedpiecewise curve to represent a time series satisfying the vertical distance betweeneach real data point and the corresponding fit curve is less than or equal to ε. Themain idea of3D-FSS is to divide a time series into segments and use quadraticfunctions to approximate segments with an error bound guarantee on each datapoint. The key to our technique is the feasible solution space (FSS) for each datapoint, where FSS of a data point is an infinite area bounded by two parallel planes.When a new data point Pjarrives, we calculate the intersection between the FSS ofPjand the intersection of the FSSs of other data points seen so far but notcompressed. If the intersection is empty, the algorithm ends the current segment,and starts a new segment from the current data point. If the intersection is not empty,the algorithm stores the latest intersection, and waits for a new data point. Thisprocess is repeated until the time series is finished.3D-FSS not only generates theminimum number of segments to approximate a time series with an error boundguarantee, but also only requires linear time O(n). Experiment results on realdatasets demonstrate that the compression performance of3D-FSS is a bit strongerthan that of GDPLA proposed in Chapter two, whereas the running time of3D-FSSis a bit more than that of GDPLA.Third, a novel decentralized and weighted fairness guaranteed datatransmission protocol WFCC is proposed. In wireless sensor networks, congestionnot only leads to packet loss, but also increases delays and reduces network throughput with a lot of energy wastage due to retransmissions. Therefore aneffective solution should be proposed to mitigate congestion to increase energyefficiency and prolong the lifetime of network. A lot of solutions are proposed tosolve this problem. However, most of them work in an open-loop controlled fashion,which may lead to the instability and rough precision of systems. To remedy thedeficiencies of previous methods, this dissertation proposes a novel decentralizedand weighted fairness guaranteed data transmission protocol (WFCC). WFCCintroduces node weight to reflect the importance of each node, and uses the ratio ofaverage packet service time to average packet interarrival time as congestion metric.Based on node weight and congestion metric, WFCC divides time axis into periodsequences, and uses closed-loop control method to mitigate congestion by adjustingthe incoming data rate of each node periodically. Importantly, WFCC firstlyintroduces a weighted fairness metric and gives its lower bound for the first time,i.e.,1(10c/9)2where0<c <0.2is a constant. Simulation results show that theweighted fairness of WFCC achieves95%on the average, which is much betterthan the existing rate-based congestion control protocol. Moreover, WFCC achieves50%and19%gains in network throughput and weighted fairness on the average,compared with PCCP that is the state-of-art rate control based congestion controlprotocol.Finally, this dissertation proposes a decentralized single neuron based datatransmission protocol SNCCP, which is a adaptive rate control algorithm based onthe combination of discrete proportional-integral-derivative (PID) control methodand single neuron. The main idea of SNCCP is to ensure that the buffer length ofeach sensor node is as close as possible to an ideal length by a feedback controlloop that adaptively calibrates the total rate of data packets entering in each sensornode periodically according to the information collected by the node recently.SNCCP define a weighted fairness metric fMand prove that fM=1O(M2), where Mis the number of periods. SNCCP has been evaluated on a real wireless sensornetwork testbed. Detailed experimental results demonstrate that the weightedfairness of SNCCP achieves99%on average, which is much stronger than theexisting rate-based congestion control protocol. Moreover, SNCCP achieves52%and18%gains on average in network throughput and weighted fairness over PCCPis the state-of-art rate control based congestion control protocol, respectively.