Research On Data Fusion Technology Of Wireless Multimedia Sensor Networks

As a new type of sensor networks, wireless multimedia sensor networks can autonomouslysensor many kinds of multimedia information, such as images, audio and video, in unattendedenvironment. By achieving much more precise performance in environment monitoring, it has abroader application prospects. However, multimedia sensor nodes have limited resource such asenergy, storage capacity and computing power, and they need to process a large amount ofmultimedia information in high computation complexity. Therefore, how to use the limited resourceof sensor nodes to obtain efficient data collection has become the most important issue in wirelessmultimedia sensor networks. The data fusion technology is an effective way to solve this problem, itcan save a large amount of network energy, enhance the accuracy and efficiency of data collectionby eliminating redundant data.After doing some in-depth research on the theoretical basis, research status and relatedtechnology about data fusion in wireless multimedia sensor networks, this thesis proposes two datafusion schemes that are suitable for wireless multimedia sensor networks. The first one is a datafusion-oriented routing protocol in multimedia sensor networks (DFRP for short) which combinedata fusion technology and routing algorithm. The innovation of the algorithm is that making use ofthe image disparity method to put the related multimedia sensor nodes into the same cluster, andcluster header which elected periodically is responsible for the efficient data fusion. The simulationresults show that, compared with another data fusion-oriented routing algorithm AFST, DFRP doesbetter in the performance of effectiveness and balance of network energy. In addition, the seconddata fusion scheme is based on the image stitch technology. Since the image stitch technology caneliminate redundant data in multiple images. The innovation of this scheme is that it takes intoaccount both nodes’ residual energy and image quality in determination of the image size whichrelated multimedia sensor nodes need to transmit. So each node will only need to send part of thesensed image in next data collection. The simulation proves that this data fusion scheme can notonly prolong the lifetime of network but also enhance the quality of collected images.