Research And Design Of The WSN Coverage Model Based On Spatial Correlation

Over the past10years, the potential application and broad prospects of WirelessSensor Network (WSN) are huge, so it has always been the hot issues in academicresearch. The significant difference between Wireless Sensor Network and traditionalcomputer communication network is the perception of information, so that we can seethe information of the physical world. The basic task of the Wireless Sensor Network isto sense and detect the specified information and event in a space. For achieving thistask the premise is the Wireless Sensor Network has to coverage the target spaceeffectively. Therefore, the coverage and detection is one of the basic and universalissues in the design and application of Wireless Sensor Networks.At present, the research about Wireless Sensor Network coverage and the detectionproblem has formed a large number of theoretical results, but in the practical applicationof Wireless Sensor Network, a large number of research results are difficult to play agood role. Most of the researches do not consider the real application scenarios andrequirements of Wireless Sensor Network. They detached the theoretical achievementsfrom application requirements. There are two main reasons for this problem:1) Most ofthe researches use the idealized assumptions such as the binary sensing model and thepure geometric computing, without considering the interference of environmental noiseand node own uncertainties. Therefore in this thesis, we will introduce the probabilitysensing model of noise considerations.2) They do not take the spatial correlationbetween the under test data and the sensor nodes into account. Adjacent sensor nodesare likely to receive the signal of the under test events, they have a certain relationshipbetween the received signal. Thus, we will introduce the spatial correlation.At first, according to the principle of spatial correlation, on the basis ofprobabilistic sensing model considering the noise and the signal propagation model inthe actual application environment, using the Neyman-Pearson theorem in statisticaldetection theory, this thesis deduces the detection threshold and the correct detectionprobability of the Mean Detector, Energy Detector, Optimal Detector, CovarianceDetector and Enhanced Covariance Detector for different data and signal characteristics. And then based on the Monte Carlo method, this thesis realizes the simulation about thesingle independent detection and the double joint detection, calculates the detectionthreshold and the probability of all the detectors. Based on the probabilistic sensingmodel, this thesis achieves the unit cover model for all the detectors, and compares andanalyses the detector performance of all the detectors. Finally, based on the unit covermodel, this thesis completes the whole area seamless coverage using the regularhexagon meshing method, and calculates the numbers of the sensor nodes to realize thewhole area seamless coverage. Experimental results show that the coverage is related tothe distance between the two sensor nodes, the smaller the distance, the betterperformance for the double joint detection. Using the regular hexagon mesh method tocover the whole area, the number of the double joint detection with the spatialcorrelation is less than the single independent detection.