A Deep-Learning-Based Self-Calibration Time-Reversal Fingerprinting Localization Approach on Wi-Fi Platform

Lili Zheng,Bin-Jie Hu,Jinguang Qiu,Manman Cui
DOI: https://doi.org/10.1109/jiot.2020.2981723
IF: 10.6
2020-08-01
IEEE Internet of Things Journal
Abstract:With the increasing demands for indoor location-based services, fingerprinting-based localization attracts considerable attention, due to that it could achieve high localization accuracy using simple equipment. However, the main problem of fingerprinting localization is that with the change of the indoor environment, the fingerprint database would be outdated, which inevitably leads to localization performance degradation. To tackle this issue, based on deep learning, this article proposes a self-calibration time-reversal (TR) fingerprinting localization approach to mitigate the effects of environmental changes without updating the fingerprint database. In the offline stage, the amplitude autoencoder (A-AE) and the phase autoencoder (P-AE) are, respectively, trained without labels to record features of the current environment. In the online stage, the trained A-AE and P-AE are used to adaptively calibrate the real-time measurements which may have been distorted due to the environmental changes. Based on the calibrated measurements, a modified TR resonating strength (TRRS) is presented for localization. The experimental results confirm the effectiveness of the proposal.
computer science, information systems,telecommunications,engineering, electrical & electronic
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: in a massive multi - user MIMO (Massive MIMO) system, how to optimize the signal processing algorithm to adapt to the complex link topologies and different power constraint conditions. Specifically, the research objective is to maximize the total rate (sum - rate) of the system by proposing a generalized multi - user massive MIMO (G4M) optimization algorithm under the conditions of satisfying the total power, the power limit of each antenna or each node. In addition, this algorithm needs to be able to handle nonlinear precoding and detection and be applicable to any number of antennas. ### Key Issues 1. **Complex Link Topology**: - Nodes in a massive MIMO system can be a single node with a large number of antennas or multiple collaborating small nodes. The collaboration methods and topologies among these nodes are very complex, and traditional MIMO system research usually does not consider these situations. 2. **Power Limitation**: - Each node participating in communication may have limited transmit power capabilities. Therefore, it is necessary to consider the power limit of each antenna or each node in the algorithm design, rather than simply the total power limit. This makes traditional algorithms based on total power limits no longer applicable. 3. **Non - linear Processing**: - The system needs to be able to handle nonlinear precoding (such as dirty paper coding DPC) and detection (such as successive interference cancellation SIC) to improve communication performance. ### Solutions The paper proposes a generalized multi - user massive MIMO (G4M) optimization algorithm. The main features of this algorithm include: - **Universality**: Applicable to any link topology, including centralized and distributed node configurations. - **Power Constraints**: Can handle total power, power limits of each antenna or each node. - **Non - linear Processing**: Supports nonlinear precoding and detection methods. - **Iterative Solution**: Gradually approaches the optimal solution by iteratively updating the transmit and receive beamforming vectors and power allocation. ### Mathematical Formulas To ensure the flexibility and universality of the algorithm, the paper introduces the Antenna Enumeration Function (AEF) to handle the antenna configurations of different nodes. Specifically, for the \(m\) - th user, the \(i\) - th row of its transmission matrix \(B_m\) represents the signal sent from the \(i\) - th antenna. Use the AEF \(u_{k,m}(q)\) to return the index of the \(q\) - th antenna of the \(k\) - th node. The power constraint formula is: \[ \sum_{q = 1}^{\hat{N}_{t,k}} \hat{b}_{u_{k,m}(q)}^\dagger \hat{b}_{u_{k,m}(q)} \leq P_k \] where \(P_k\) is the maximum allowable transmit power of the \(k\) - th node, and \(\hat{N}_{t,k}\) is the number of antennas of the \(k\) - th node. ### Summary This paper solves the complex link topology and power limitation problems in the massive MIMO system by proposing the G4M algorithm, providing a flexible and efficient optimization method that can maximize the total rate of the system in multiple application scenarios.