An Improved DV-Hop Localization Algorithm Based On Differential Evolution

Wireless Sensor Networks(WSNs) consist of many inexpensive sensor nodes with tiny size. When sensor nodes are scattered into a sensor field, they will wirelessly communicate with each other and do networking by self-organization, in order to gather data from the sensor field. The development of Internet of Things(IoT) is an inevitable trend in the present information society，as the core technology of IoT, WSNs tightly couple the physical world with the InfoWorld. Localization is one of the key technology of WSNs, since many data must be monitored rely on location information, it is significant to do research on localization algorithm.Localization algorithm can be divided into range-based and range-free on the basis of the need for measuring distance. The high position accuracy can be obtained using range-based localization algorithm, however, it increase the system costs with the additional ranging hardware. Range-free localization algorithm has lower system costs and energy consumption than range-based algorithm, and its position accuracy can meet most localization requirement as well, so it becomes a research hotspot localization technology. DV-Hop is a typical range-free localization algorithm in good performance, but it has a defect of unsatisfactory position accuracy.In this paper, I made a simulation for DV-Hop algorithm, and analyzed its error sources, which can be divided into internal error and external error. The external error are environment-related errors, which can’t be avoided by improving the algorithm. Internal errors of DV-Hop may come from two stages: the stage of estimating the hop distance and the stage of localization. Then I proposed an improved DV-Hop localization algorithm based on differential evolution called DEPC-Hop algorithm. Comparing with the existing algorithm, DEPC-Hop algorithm has three improvements:1. In DV-Hop algorithm, the average hop distance of beacon nodes can’t reflect the overall conditions of network, so I proposed a self-adaptive estimating method for average hop distance to solve the problem. First, unknown nodes receive the average hop distance form the nearest beacon node, when calculating the distance between the unknown node and other beacon nodes, I use the self-estimating method to adjust the average hop distance by different beacon nodes, then calculate the hop distance by the new average hop distance. Thus we can get a estimation agrees better with the reality.2. Aiming at the problem that the existed localization models have some defects, I proposed a localization model based on improved Differential Evolution according to the characteristics of DV-Hop algorithm. To decrease the degree of variation, I selected the neighbour individuals to form the differential vector in the first stage. In variation stage, the scalar number F would attenuate with the increase in generation to adapt the small-scale searching. When selected the population size, each unknown node has three corresponding individuals, whose centroid would be regarded as the position of the unknown node after the evolution.3. A position-correction algorithm is proposed to solve the problem that DE algorithm is easily trapped in the local optimum. When the initial localization finished, each beacon node would amend the result according to the hop distance and position relation between the unknown node and the beacon node, this method can further improve the accuracy.DEPC-Hop algorithm has been tested in simulated environment by MATLAB2012b. Performance analysis and simulation results showed that DEPC-Hop algorithm has faster convergence, compared with DV-Hop algorithm, the localization accuracy can be enhanced by10.22%.