Currently, indoor positioning is not as widely used in daily life as outdoor positioning, and fewer service types are provided. However, if accurate indoor positioning services can be provided, outdoor positioning services can be extended indoors. At the same time, if indoor positioning technology can be combined with other technologies such as physiological information acquisition technology, then home medical care or home lifestyle identification can be performed to achieve the goal of ubiquitous care. For example, indoor positioning technology can determine a patient’s location, such as in the living room, bedroom or bathroom. By combining indoor positioning technology with physiological information, the patient’s behavior, such as watching TV on the sofa, going to bed in the bedroom, or staying in the bathroom for overly long, can be speculated. The patient’s behavior can determine whether there is a possibility of an accident and provide appropriate assistance.

Furthermore, indoor positioning systems can provide the location of an automatically classified product to manufacturers to determine whether the product is classified correctly. Employees’ current locations can also be provided to managers through indoor positioning services to determine whether or not the employees have entered restricted areas and violated safety regulations. Other services such as indoor environment monitoring, indoor disaster protection, rescue management [1,2], can all be accomplished with the advancement of indoor positioning technology.

Since indoor environments may be quite complicated, indoor positioning systems may encounter difficulties that affect the accuracy, such as the attenuation of signal strength due to different obstructions, signal drift or positioning algorithm design. In this work, the strength of the Bluetooth wireless signal is used to calculate the person’s location. To obtain a person’s position accurately and rapidly, the system needs to overcome interference signal sources in the environment, such as the electromagnetic interferences generated by metal and electronic products, which attenuates the signal strength and reduces positioning accuracy.

This work uses artificial intelligence to build a novel indoor positioning framework and achieve accurate personnel positioning in a complex indoor environment. The proposed model is used to process the signals and complete the establishment of the algorithm.

There are many different positioning methods for positioning systems. For wireless signal positioning, multiple fixed communication receiving points are set up to communicate with the measured object, and the relative distance is calculated. Many approaches have been proposed to calculate the relative distance as well as the accuracy and computational complexity. Several papers have proposed different methods of indoor personnel positioning.

In 2013, Tsuboi et al. [3] proposed the Ultra-High Frequency Radio Frequency Identification (UHF RFID) based personnel positioning system with a path tracking function. The authors proposed that the RFID method accurately and quickly calculates the personnel’s position and records the personnel’s historical position through the database. The author installed several passive RFID tags on the ceiling of the experimental environment to form an RFID tag array. The subject holds a handheld reader. When the subject moves, the reader reads the RFID tag and sends the tag number to the database to record the historical path and location. This method has very low complexity in positioning calculations and implements localization quickly and accurately. However, it requires a large number of RFID tag installations.

In 2015, Schafermeyer et al. [4] proposed a method that uses simple wall-mounted radio frequency transceiver and IR sensors capable of fingerprint identifications for individual identification. The method achieved a classification accuracy of 98% when using a Gaussian mixture model for classification. In the same year, Ma et al. [5] proposed a method to identify personnel entering or leaving a building. The method used Wi-Fi signals and combined the One-Class SVM and Binary-Class SVM classifiers.

In 2016, Kim et al. [6] proposed a personnel location algorithm based on the wireless signal strength received by a mobile phone. The authors used the K-Nearest Neighbor classification (KNN) machine-learning algorithm to establish a positioning model. The KNN model was used to group the positions of people according to the strength of the received signal. The positioning error of this system was 0.33 to 8.67 m, and the accuracy varied significantly in environments with many interferences.

In 2017, Alkandari et al. [7] proposed a method to estimate the speed between two fixed points based on device-free passive localization (DfPL) [8]. The authors proposed using one access point and onemonitoring point to detect DfPL and estimate the personnel’s indoor position. In the same year, Oguntala et al. [9] proposed using the received signal strength of passive RFID to measure the distance for personnel localization. The system used particle filters to analyze RSS to localize the target position in an indoor environment.

In 2018, Cui et al. [10] proposed a ZigBee-based personnel positioning system. The authors proposed a Zigbee railway tunnel employee positioning system to enable tunnel managers to grasp the exact location of the tunnel personnel in real-time, achieve appropriate deployment and effective command of employees, as well as improve management and emergency command capabilities. The authors set up four Zigbee fixed nodes in the tunnel and used the K-means machine learning method to establish a model. When the fixed node receives the signal sent by the mobile node worn by the personnel, the signal strength calculated based on the received signal is input into the model to calculate the current location of the personnel. The method used machine learning to establish a model with an accuracy of 95% when the personnel’s error is within 2 m.

Al-Khaleefa et al. [11] proposed the maximum feature adaptive online sequential extreme learning machine (MFA-OSELM) in 2019 to process the cyclic dynamic factors due to a person’s movements in an internal environment. In the same year, Xue et al. [12] proposed a WiFi-based localization model by modifying the large localization errors and enhancing the Gaussian process regression (MEGPR). The model increased the effectiveness and feasibility of accurate indoor positioning. However, it required a multitude of access points to provide Wi-Fi signals, thus increasing the implementation cost.

The MLP is a kind of feedforward neural network. It has a three-layered structure: an input layer, a hidden layer and an output layer. The input layer is the current material to be learned. The hidden layer is the feature node that needs to be learned, and each node is a neuron with a nonlinear activation function so that its output meets the required results. The output layer is the category that needs to be learned. Each layer is a fully-connected layer in the middle. The multi-layer perceptron was widely used in the 1980s in problems such as speech recognition, image recognition, machine translation and has achieved good results in classification problems.

Early neural network architectures have no time concept and no label learning for sequence signals. In 1982, Hopfield [13] proposed a Hopfield neural network with 2n states, where the value of each neuron is either 1 or 0, and the output is a four-bit binary number. There are 16 network states in total due to the network’s recursive characteristics. If the network is stable, it converges from any initial state to a stable state; otherwise, it does not diverge because each neuron has only two states.

Jordan [14] proposed a structure in 1986 where the learning method includes labeling the sequence signal, and the output provides feedback to hidden nodes, as shown in the figure. First, the network multiplies the input and the weight values of a neural network and adds on the deviation value of the node. Then, the value that has been calculated once and the output values are recalculated.

Since the RNN is modeled by serial signals, it achieves better results if the signals are correlated before and after. Therefore, it is often applied to natural language, handwriting recognition, weather, or sensor values.

However, as the network’s hidden layer becomes deeper and deeper, the gradient vanishing or gradient exploding problem may occur, making it difficult to use in situations where the signal time is long.

The activation function in neural networks mainly uses nonlinear equations to solve nonlinear problems. Without activation functions, the neural network is a linear combination of operations. Since both the hidden and the output layers are input results of the upper layer, they are calculated with the weight value and the deviation. The calculation result is regarded as the layer’s output so that the output and the input have only a linear relationship. The learned model cannot solve nonlinear problems without using a nonlinear activation function. In this work, we have provided four different activation functions when learning the framework. Activation functions with good outcomes are selected through experiments to establish a fall detection model.

The system architecture diagram is shown in Fig. 5. The architecture of the proposed system incorporates four hardware devices: wearable devices worn by the user, information receivers for receiving the packets, routers for establishing the local networks and management computers with interface programs.

The system continuously sends information packets to the information receiving device through the Bluetooth broadcast mode. The information receiver first uses the algorithm to calculate the RSSI value of the packet and sends the user’s positioning information to the computer through the wired network. The management computers first analyze the user’s status information in the packet, calculate the current user location through the positioning algorithm, display the user’s location in the management interface program and finally upload the positioning information to the cloud database for storage.

The device uses Texas Instruments’ CC2541 Bluetooth wireless chip, the core of which is an 8051 microcontroller. The power source is a 3.7 V 1000 mAh polymer lithium battery with a working endurance of about one month. The components of the wearable device are shown in Fig. 6.

The embedded development software of the wearable device is the IAR Embedded Workbench F, and the program is developed using the C language.

For the system to have low power consumption, automatic detection of power and increased stability, three functions have been added: a low power consumption mode, ADC voltage detection and a watchdog timer (WDT). The low power consumption mode switches to sleep mode after completing an event to reduce power consumption. The ADC voltage detection checks whether or not the current battery power is sufficient when the system is turned on. The WDT checks whether or not the program continues to operate upon the completion of an event. The flowchart of the wearable device system is shown in Fig. 7.

The flowchart of the indoor positioning system is shown in Fig. 10. The receiver receives the packet sent by the wearable device and discards the packet information not sent by the device. The receiver’s internal algorithm then calculates the RSSI value of the received packet. Next, the RSSI value of the packet is sent to the computer interface program via the TCP/IP socket. Finally, the computer interface program uses a positioning algorithm to calculate the user’s location and marks the location on the computer interface program. The management personnel monitors all users’ locations in the field through the interface program. When a user accidentally falls or activates the emergency button, the displayed user mark will be marked in different colors so that the management personnel can quickly proceed to the location and deal with the situation.

The flowchart of the management interface program system used by the management personnel is shown in Fig. 11. The management personnel can check the user’s position at any time through the management interface program to ensure the user’s safety. After the user installs the wearable device, the device must be paired with the transmitter device through the management interface program. When a user enters the detection area, the user’s location and status can be read on the management interface program. When the user leaves the detection area, the user must use the management interface program to unpair the wearable device and retrieve the device for later use.

The information receiver uses the Raspberry Pi 3B as its development platform. Due to the device’s small size, strong computing power and multiple functions, the proposed system adds a USB Bluetooth receiver to its microcomputer and uses the C language to develop a packet receiving program. This program’s main functions are to receive information packets from the wearable device, confirm that the packet information is correct, calculate the RSSI value of the packet, reassemble the packet information, and send it to the computer. The flowchart of the information receiver program is shown in Fig. 12. The information receiver automatically finds and receives the packet information broadcasted by the wearable device, recombines the device number, the user status and the RSSI value of the packet information into a new information packet. Finally, it sends the new packet to the interface program for processing using the TCP/IP Socket.

The system uses artificial intelligence to establish a positioning model and finds an optimal algorithm to solve the environmental factors that affect indoor positioning. Spyder and Python are used as the development software to create the model. The system uses Tensorflow to build a model and various neural network models to find a high-accuracy model. The established model is reproduced in C/C++, and the algorithm is applied to the indoor positioning system.

There are two ways to place the information receivers, using either four or six receivers and dividing the experimental field into thirty-two areas. The receivers are placed in the experimental field, as shown in Fig. 16.

The locations of the receivers are at 1, 3, 4 and 6 for the placement with four information receivers and at 1, 2, 3, 4, 5, and 6 for the placement with six information receivers. The training materials also adapt to the thirty-two-area receiver signals based on these two placement methods. Each area obtains one hundred seventy-five to one hundred and eighty-two data values, and a total of five thousand seven hundred thirty-two signals are acquired. The signal values of thirty-two areas are shown in Fig. 17.

The model is divided into two structures, with four or six information receivers, to select the optimal number for information receivers placement. The wearable device is updated once per second, and the calculations are performed by a computer. As such, the architecture is expected to consume a lot of resources.