Wireless Sensor Network Distributed Multidimensional Scaling Location Algorithm

Wireless sensor networks combine sensor technology, embedded systems, Internet and wireless communication, distributed information processing technology. It has wide application prospects in military affairs, environment monitoring, disaster relief and other commercial field. The location of the node in wireless sensor network is random and unknown. How to achieve the location of the node accurately is the prerequisite and hotspots in the research of the sensor network applications.Multidimensional scaling technology series is a kind of efficient node localization method. However, there are a certain limitations in this method. It requires dissimilarity between entities and the distance of them to maintain the linear relationship. But the non-metric multidimensional scaling technique in positioning of this relationship doesn't have so strict requirement. It only needs the hierarchy of monotonicity in order. Currently the using of non-metric multidimensional scaling techniques still has certain technical difficulties. The cost of computing and communications will be increased as the number of nodes in the network increases rapidly. It's not suitable for large-scale wireless sensor networks, and in some cases the adaptability in the movement of parts nodes. If the network node in a small amount of movement, need to re-locate all the nodes. If the movement in the network in a small amount of nodes, it needs to re-locate all the nodes.The positioning mode of the multidimensional scaling technique in wireless sensor networks is discussed. A kind of algorithm, distributed multi-dimensional positioning algorithm for scaling is proposed, which is called NMDS-RSSI-C(D)(Non-metric Multidimensional-Receive Singnal Strength Indicator-Cluster(Distributed)). In this algorithm the strength indicator values of the wireless signal between nodes will be the dissimilarity of data between nodes directly. The cluster head in each local network computes the shortest path between the node within the local network. Using non-metric multidimensional scaling techniques to calculate the coordinates of the nodes. We calculate the local relative coordinate map by using local information, and then get a global relative coordinate map by combining a number of local relatived. Non-metric multidimensional scaling technique asks the relationship between the dissimilarity and distance of entities in the order of just a monotonous sequence of levels. It does not need to be expressed quantitatively. Using the strength value of the signal received by the nodes to define RSSI ranging. Then the wireless signal propagation model convert that value to distance. Combine RSSI and NMDS can locate the wireless signal strength values directly and lessen the complex transform process, reduce the calculation error because of the inaccuratel parameters in the wireless signal propagation model.Finally, we compare the algorithm of NMDS-RSSI-C (D) and the MDS-MAP (D). When it's in low node connectivity the positioning error are both quite large in the two method. But the algorithm we mentioned in this article get less average position error in the same nodes connectivity. In this algorithm, the larger connectivity of the nodes the less of the combination times, the error comes from the combination will be reduced. Simulation results show that for large wireless sensor networks, the algorithm can achieve effective positioning and the error is quite small.