Abstract:Wireless localization is essential for tracking objects in indoor environments. Internet of Things (IoT) enables localization through its diverse wireless communication protocols. In this paper, a hybrid section-based indoor localization method using a developed Radio Frequency Identification (RFID) tracking device and multiple IoT wireless technologies is proposed. In order to reduce the cost of the RFID tags, the tags are installed only on the borders of each section. The RFID tracking device identifies the section, and the proposed wireless hybrid method finds the location of the object inside the section. The proposed hybrid method is analytically driven by linear location estimates obtained from different IoT wireless technologies. The experimental results using developed RFID tracking device and RSSI-based localization for Bluetooth, WiFi and ZigBee technologies verifies the analytical results.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to achieve more accurate object localization in an indoor environment. Specifically, the paper proposes an indoor localization method based on hybrid wireless data fusion, which combines the developed radio - frequency identification (RFID) tracking devices and multiple Internet of Things (IoT) wireless technologies. The main objective is to improve the localization accuracy by reducing the cost of RFID tags (i.e., installing RFID tags only on the boundaries of each area) while utilizing the received signal strength indication (RSSI) of multiple wireless technologies such as Bluetooth, WiFi, and ZigBee.
### Main Problems
1. **Challenges in Indoor Localization**: Traditional commercial GPS systems are unable to provide accurate indoor localization information due to the multipath effect and the path loss when signals pass through different building materials.
2. **Balance between Cost and Accuracy**: How to improve the accuracy of indoor localization systems while reducing costs.
3. **Multi - sensor Data Fusion**: How to effectively combine the data of multiple wireless technologies to improve the accuracy and reliability of localization.
### Solutions
The paper proposes a segmented hybrid indoor localization method. The specific steps are as follows:
1. **Use of RFID Tags**: Install RFID tags on the boundaries of each area, and RFID tracking devices are used to identify the area where the object is located.
2. **Fusion of Multiple Wireless Technologies**: After determining the area where the object is located, use the RSSI information of Bluetooth, WiFi, and ZigBee for more accurate localization.
3. **Linear Combination Estimation**: Minimize the error distance between the final estimated position and the true position by linearly combining the estimated values of different localization techniques.
### Mathematical Model
The paper proposes an optimization problem to find the optimal weight coefficients by minimizing the generalized distance error (such as the mean - squared error MSE and the mean absolute error MAE). The specific formulas are as follows:
- **Generalized Distance Function**:
\[
d_V(P, Q)=\|P - Q\|_V = V(P - Q)=V_x(x_p - x_q)+V_y(y_p - y_q)+V_z(z_p - z_q)
\]
where \(V_u(t)\) is a strictly convex function.
- **Linear Combination Estimation**:
\[
\hat{P}(P)=\left(\sum_{i = 1}^N\alpha_{x,i}\hat{x}_i,\sum_{i = 1}^N\alpha_{y,i}\hat{y}_i,\sum_{i = 1}^N\alpha_{z,i}\hat{z}_i\right)
\]
Constraints:
\[
\sum_{i = 1}^N\alpha_{u,i}=1,\quad\forall u\in\{x, y, z\}
\]
\[
\alpha_{u,i}\geq0,\quad\forall i\in\{1,\ldots,N\},\forall u\in\{x, y, z\}
\]
- **Optimization Problem**:
\[
\alpha^*=\arg\min_{\alpha\in A}\frac{1}{M}\sum_{j = 1}^M d_V(\hat{P}(P_j),P_j)
\]
\[
=\arg\min_{\alpha\in A}\sum_{u\in\{x, y, z\}}\sum_{j = 1}^M V_j^u\left(\sum_{i = 1}^N\alpha_{u,i}\hat{u}_j^i - u_j\right)
\]
### Experimental Results
The experimental results show that the proposed hybrid localization method outperforms the localization methods using Bluetooth, WiFi, or ZigBee alone in terms of both mean - squared error (MSE) and mean absolute error (MAE). In particular, the method combined with RFID tags improves the localization accuracy by 66%, 79%, and 74% within a 60 - meter range (relative to Bluetooth, WiFi, and ZigBee).