Sai Prasanth Kothuri

Wanna do some puzzles? Give it a try here 🤓🤓

Solve Puzzle

Deep Neural Networks for

Localisation in 6G

Final_ICMLCN_poster_sai_prasanth.pdf

Poster Presentation at ICMLCN Conference, Stockholm, Sweden.

Localisation in 6G using Deep Learning Techniques

In the rapidly advancing domain of wireless communication, enhanced user positioning is becoming increasingly critical. Highly precise positioning is essential for applications such as autonomous vehicles, digital twins, augmented and virtual reality, and the Industrial Internet of Things (IIoT). My research aims to develop new algorithms to achieve centimeter-level positioning accuracy in Indoor Factory scenarios which are NLOS dominant in nature due to high clutter density. This centimetre accuarcy can be achieved by leveraging Artificial Intelligence/Machine Learning (AI/ML) techniques using the real-world datasets from our Indigenous 5G Testbed at IIT Madras.

Localisation using cellular networks has traditionally been a challenging problem, with current accuracies typically in the range of tens of meters. However, with the advent of AI/ML and high-frequency phased arrays, new research avenues have opened up, targeting centimeter-level accuracies. Such precise positioning would enable a wide array of use cases, particularly in indoor and industrial environments.

Types of Localisation Techniques used in 5G-NR

Positioning methods in 5G communication are crucial for various applications ranging from navigation and location-based services to enhanced network management and emergency services. The number of Base Transceiver Stations (BTS) required and the algorithms used for positioning depend on the specific method employed. Here’s a detailed breakdown of these requirements and the associated algorithms for each positioning method:

Global Navigation Satellite System (GNSS)

BTS Requirement: GNSS does not rely on BTS for positioning. Instead, it requires signals from a minimum of 4 satellites to calculate a 3D position (latitude, longitude, and altitude).
Position Calculation Algorithms:
- Trilateration: Determines the position by measuring the distance from multiple satellites.
- Kalman Filter: Used to estimate and smooth the position over time.

Cellular-Based Positioning

BTS Requirement:
- Cell ID: Only one BTS is required, providing coarse location based on the cell coverage area.
- TOA, TDOA, AOA: Requires signals from at least 3 BTS for triangulation and accurate position estimation.
Position Calculation Algorithms:
- Time of Arrival (TOA): Calculates distance based on the time it takes for a signal to travel from the BTS to the device.
- Time Difference of Arrival (TDOA): Uses the difference in arrival times at different BTS.
- Angle of Arrival (AOA): Estimates the angle of the incoming signal at multiple BTS to triangulate the position.
- MUSIC and ESPRIT: Advanced algorithms for high-resolution estimation of signal angles (used in AOA).

Wi-Fi Positioning System (WPS)

BTS Requirement: Typically requires 3 or more Wi-Fi access points to triangulate the position for accurate location determination.
Position Calculation Algorithms:
- Fingerprinting: Matches real-time signal strength measurements with a pre-constructed database of known locations.
- Triangulation/Trilateration: Similar to cellular methods but using Wi-Fi signal strength.

Bluetooth Low Energy (BLE) Beacons

BTS Requirement: Requires multiple beacons (typically 3 or more) for accurate indoor positioning.
Position Calculation Algorithms:
- RSSI (Received Signal Strength Indicator): Uses the strength of the signal received from the beacons to estimate distance.
- Proximity Detection: Determines the nearest beacon for proximity-based applications.
- Fingerprinting: Similar to Wi-Fi, using signal strength patterns for location estimation.

Ultrawideband (UWB) Positioning

BTS Requirement: Generally needs at least 3 UWB anchors (fixed transmitters) for accurate positioning.
Position Calculation Algorithms:
- Time of Flight (ToF): Measures the time it takes for the UWB signal to travel from the device to the anchor.
- TDOA: Similar to cellular TDOA but using UWB signals for higher accuracy.

Page updated

Report abuse