Research On RSSI-based Regional Positioning Technology

With the rapid development of ubiquitous computing,wireless sensor networks and the Internet of Things(IOT),indoor area positioning technology plays an important role in intrusion detection,elderly health monitoring,smart buildings and other fields.Because of the non-line-of-sight,wide coverage,and all-weather work of wireless signals,the area positioning based on received signal strength indication(RSSI)of wireless signals is superior to the traditional indoor area positioning methods based on infrared,video,and optical perception..Domestic and international indoor area positioning technologies based on wireless signal RSSI have made important progress.However,the existing methods mainly have the following three problems in the actual application process: First,the dynamic and unpredictability of the radio channel effect(Such as multipath propagation,channel fading,etc.)the impact of interference on RSSI.Second,because Bluetooth low energy(BLE)beacons only allocate three broadcast channels,when more than three beacon stations are installed near the receiver,the beacon signal cannot be restored correctly due to collisions.Third,the transmission power of the BLE beacon is low and the transmission range is narrow.Therefore,the receiver is more likely to be located in the boundary area of the BLE beacon station.At the same time,BLE beacon stations are usually powered by batteries,so there will be a signal vacuum when the batteries are exhausted.For problem one,this paper builds a model between RSSI and physical location information between Bluetooth anchor nodes based on machine learning and deep learning algorithms.This model reduces the channel change of RSSI sequence by capturing the correspondence between coarse-grained features and positioning regions Dependence,first,using convolutional neural network(CNN)to learn and capture the features of RSSI sequence to extract the fine-grained features of the center point of the region.Then,using the storage and memory characteristics of bidirectional long short-term memory(Bi LSTM)to learn the coarse-grained features of the hidden area range in the current and past RSSI sequences.Finally,by using the attention mechanism,the mapping relationship between the features of the RSSI sequence and the location of the region is established by fusing the features of coarse and fine granularity to obtain the location information of the region.To solve the second problem,this paper establishes a Bluetooth Mesh network and establishes many-to-many communication of wireless devices,so that many nodes can receive messages through relays,expanding the transmission distance of the network,increasing the number of available nodes,reducing network conflicts,and enhancing The stability of the network,in addition,when the network is running,there is no need to create and manage complex routing tables,route discovery tables,etc.to transfer information,saving the memory space required to maintain the network operation;for question three,the formation of Mesh The network flooding message transmission through the multi-path(Multi path,MP)enables information to reach the target node through multiple paths,expanding the transmission range,and the network transmission will not cause fatal effects due to the failure of any node To ensure the stability of the network.The experimental results of regional positioning in real indoor environment show that,compared with the current grid area comprehensive probabilistic positioning model with the best positioning effect,the proposed method reduces the computational complexity while improving the accuracy of regional positioning and adapting to the environment ability.