Abstract
Localization is the basis of wireless sensor networks application and technical research, but in the case of anchor sparse nodes, there are often problems of low localization accuracy and location misjudgment. When there is no neighbor reference anchor node in the communication range of a blind node, it is often abandoned and not be located. Multi-stage localization is selected to solve this problem, blind nodes successfully located in the previous stage are used as cooperative anchor nodes in the next stage to increase the density of anchor nodes while maximizing the number of locating nodes. For blind nodes, their locations may be misjudged when the neighbor reference anchor nodes are distributed along or close to a straight line. Bounding-box is used to define blind node search boundary boxes, fuzzily locate blind nodes instead of direct locating, and more accurate node locations are found through the intelligent algorithms. At the same time, the krill herd algorithm based on random search is used to find the location of blind nodes by moving krill individuals. Experimental results show that the multi-stage min-max krill herd(MKH) proposed in this paper has higher localization accuracy and localization stability under the same conditions.