Research On RSSI Indoor Wireless Location Method Based On ZigBee

Among the existing indoor wireless methods,the positioning method of Received Signal Strength Indication(RSSI)based on Zig Bee is widely used in the field of indoor wireless positioning due to its advantages of low power consumption,low cost and easy implementation.However,the existing RSSI positioning methods have a low utilization rate of the characteristic information of RSSI data,resulting in the positioning accuracy of the Zig Bee-based RSSI positioning method is generally not high.To solve this problem,this paper uses machine learning algorithm to build a feature extraction model,and makes full use of the time correlation characteristics and spatial distribution characteristics of RSSI data for positioning prediction,overcoming the defects of traditional RSSI positioning methods that only consider signal strength and other single features,so as to improve positioning accuracy.In the actual positioning stage,the positioning method based on RSSI can be divided into ranging model positioning and location fingerprint positioning according to the different application methods.In this paper,the two positioning methods are analyzed and improved to improve the positioning accuracy.First of all,for the traditional RSSI ranging model establishment method only uses a large number of RSSI data of a single moment for fitting,while ignoring the time correlation characteristics of the RSSI data itself,resulting in low accuracy of the ranging model.This paper uses Gated Recurrent Neural Network,The Sparrow Search Algorithm(SSA)has the advantages of simple and efficient time series processing ability and nonlinear approximation ability,and a Sparrow search algorithm optimization GRU ranging model establishment method is proposed.By structuring RSSI data into sequence data as GRU input for feature learning,making full use of the time characteristic information of RSSI data to establish the nonlinear mapping relationship between RSSI sequence data and target distance,and using sparrow algorithm to optimize the super parameters in GRU training process,reducing the influence of artificial setting of key training parameters on model performance.Further improve the ranging accuracy of the ranging model.In the positioning stage,after converting the RSSI value between each beacon node and the target node into the corresponding distance value through the trained ranging model,the positioning solution can be carried out according to the least square method.Secondly,the existing RSSI location fingerprint positioning methods only use the single features of location fingerprint data,such as spatial features and time features,for positioning,and the model training parameters need to be manually specified,leading to the deficiency of low positioning accuracy.In this paper,a Convolutional Neural Network(CNN)based on the sparrow algorithm is proposed,which combines the spatial feature extraction ability of CNN with the time series processing ability of GRU.In order to improve the positioning accuracy,the spatial and temporal characteristics of location fingerprint data are fully used for positioning.In the training process,the sparrow search algorithm is used to optimize the key parameters in the GRU network training process,which reduces the influence of artificial setting of key training parameters on the model positioning effect and further improves the positioning accuracy of the model.Experimental verification shows that the two improved RSSI methods proposed in this paper can effectively improve the positioning accuracy and meet the positioning requirements.