Abstract:We propose and demonstrate accurate 3D real-time indoor localization via broadband nonlinear backscatter in passive devices. The proposed method does not need any relative motion between a reader and a tag or the use of reference anchor nodes. In the conventional radio frequency identification (RFID) system, a passive tag responds to a reader by switching its antenna "on" and "off". The operation of such conventional backscatter is essentially "linear", since the reader-to-tag (downlink) and tag-to-reader (uplink) signals overlap on the same carrier frequency. Although linear backscatter is straightforward, the self-jamming problem caused by strong leakage signals from the transmitter to the receiver is notorious and poses many constraints on the received signal quality, operation bandwidth, modulation flexibility and system complexity. To enable high accuracy real-time 3D indoor localization for passive devices, we show that nonlinear backscatter is more effective than linear backscatter. Nonlinear backscatter exploits nonlinear elements in passive devices to generate second or higher-order harmonics as the uplink response. Separation of downlink and uplink on different carriers allows immediate self-jamming cancellation and direct un-modulated carrier phase decoding, hence resulting in better received signal quality and broad bandwidth of operation, both of which are critical for the localization system. The broad bandwidth allows the design of an efficient phase-based ranging algorithm - heuristic multi-frequency continuous wave (HMFCW) ranging which resolves ambiguous phase cycles with heuristically optimized sparse carrier frequencies. HMFCW ranging can correctly pin down the phase cycle integer with 100% reliability as long as the phase error falls within 90 x BW% (percentage bandwidth). In our present implementation, we achieved a median ranging error below 1 cm under phase error bounds of 50. We realized real-time 3D localization from differential ranging by nonlinear conjugate gradient (CG) search for hyperboloids intersection in a multi-static transceiving system with 1 Tx antenna and 4 Rx antennas. The measured 3D localization median er-ror was 3.5 cm in the indoor environment. Presently, the measurement latency was less than 0.155 seconds. We will present system design, algorithms and a prototype with experimental evaluation.

A Structured Bidirectional LSTM Deep Learning Method for 3D Terahertz Indoor Localization

SIABR: A Structured Intra-Attention Bidirectional Recurrent Deep Learning Method for Ultra-Accurate Terahertz Indoor Localization

3D LiDAR-Based Global Localization Using Siamese Neural Network

Sparse Bayesian Inference Based Direct Localization for Massive MIMO.

Indoor Localization Using Bidirectional LSTM Networks

Three-Dimensional Indoor Localization and Tracking for Mobile Target Based on WiFi Sensing

A 3D mobile positioning method based on deep learning for hospital applications

A Three-Dimensional Wireless Indoor Localization System

Toward Long-Term Effective and Robust Device-Free Indoor Localization via Channel State Information

A Channel Adaptive WiFi Indoor Localization Method Based on Deep Learning

3d Real-Time Indoor Localization Via Broadband Nonlinear Backscatter In Passive Devices With Centimeter Precision

Learning to Localize with Attention: from sparse mmWave channel estimates from a single BS to high accuracy 3D location

Deep Learning for Ultra-Wideband Indoor Positioning

Discrete Indoor Three-Dimensional Localization System Based on Neural Networks Using Visible Light Communication

Hierarchical Stage-Wise Training of Linked Deep Neural Networks for Multi-Building and Multi-Floor Indoor Localization Based on Wi-Fi RSSI Fingerprinting

Recurrent Neural Networks For Accurate RSSI Indoor Localization

Spotlight: A 3D Indoor Localization System in Wireless Sensor Networks Based on Orientation and RSSI Measurements

Attention Mechanism and LSTM Network for Fingerprint-Based Indoor Location System

UWB Indoor Localization Using Deep Learning LSTM Networks

Hierarchical Multi-Building And Multi-Floor Indoor Localization Based On Recurrent Neural Networks

Decimeter Level Indoor Localization Using Hybrid Measurements of a Distributed Single Receiver