Research On Gossip Algorithms For Distributed Blind Zones Recognition

At present, the blind zone detection of mobile communication system is mainly implemented by the professional and specialized technology tool, which has long testing period, high cost and lack of flexibility. In order to solve this problem, we can deploy a large number of wireless sensor nodes in the monitoring area, using the distributed compressed sensing and scattered data surface reconstruction technology to detect the two-dimensional distributed blind area. Finally, one can get the result of blind zone detection as long as access to any node in the network.Firstly, we apply geometrical optics theory to simple obstacle shadow model, and we combine that model with the path loss model and 6 diameter multipath Rayleigh fading model. According to the International Telecommunication Union ITU-R m.1225 standard, we establish network monitoring area of wireless signal propagation model and use Kalman filter for filtering the random noise and some small scale fading of the signal. Then we introduce the classical compressed sensing algorithm of three major steps, as well as the basic theory of gossip algorithm in detail. We apply Eavesdropping-Based Gossip Algorithms which has faster convergence speed and is able to resist packet loss to the distributed compressed sensing algorithm. Its efficiency and feasibility have been proved by comparing and analyzing simulated results. In addition, in order to know the number of nodes which play a part in reconstructing signal with given accuracy, this paper research the relationship between margin of error and the number of nodes k. The result guarantees the theoretical performance.After using wireless sensor nodes getting its local received signal strength, we apply distributed compressed sensing based on Eavesdropping-Based Gossip Algorithms so that each node in the network knows the distribution of the blind zone. But the received signal strength distribution known by nodes is discrete. Signal distribution in the entire space is not complete, and we cannot directly get the location of the blind zone. Thus, it is necessary to use these discrete values to reconstruct the whole spatial signal coverage through the scattered data surface reconstruction technique. We study the effect of different schemes based on interpolation in the RSSI map reconstruction, and put forward the improvement scheme which combine DBSCAN algorithm and convex hull algorithm according to the characteristics of the system model. Through the simulation results, it can be proved that the combined method is suitable to this paper scene, and can greatly improve the error tolerance of the system. It has very high practical value.As mentioned above, in this paper, with applying the integrated use of graph theory, distributed signal processing, optimization theory and linear filter theory of technology and so on, we propose a two-dimensional distributed blind zone detection algorithm. This algorithm fills in the gaps in this field at home and abroad and has strong practicability and frontier.