Positioning in Cellular Networks

I lead the development of AI/ML models and advanced signal-processing algorithms for wireless positioning, spanning delay-, angle-, carrier-phase-, and hybrid-based positioning across diverse deployment scenarios. I have contributed to 3GPP RAN1/RAN2 positioning standardization from Releases 16 through 19, and now support teams in translating these innovations from standards and algorithms into real-world, field-deployable positioning solutions.

My work combines classical signal processing with AI/ML-driven positioning techniques. It includes angle-based position estimation using downlink AoD and uplink AoA in massive MIMO-OFDM systems; ESPRIT-based time-of-arrival estimation and multilateration across multi-cell deployments; and sidelink positioning for V2X scenarios. On the AI/ML side, I work on CNN-Transformer architectures for NLOS and indoor localization, along with low-overhead AI/ML techniques that reduce the signaling, measurement, and data-collection overhead associated with positioning procedures.

This journey began at CEWiT, IIT Madras, where I initiated activities around 3GPP wireless-positioning standardization, developed a 3GPP-compliant positioning system simulator, and helped initiate work on RedCap and sidelink positioning through Releases 16 and 17. Today, at Tejas Networks, I lead Release 20 activities on AI/ML-enabled positioning, drive the associated intellectual- property and patent portfolio, and work with engineering teams to translate positioning algorithms and standards concepts into practical, field-deployable solutions.

Selected publications

• PosNet: A CNN-based Transformer Model for UE Localization in NLOS Dominated Scenarios — IEEE Communications Letters (submitted 2026). Transformer-backed CNN that attends to the most informative CIR taps, reaching 5.4 cm accuracy for 90% of UEs in NLOS-dominated channels.
• A Hybrid CNN-Transformer Framework for Precise Indoor Positioning in 6G Networks — IEEE MeditCom 2026. Hybrid CNN-Transformer architecture pairing local feature extraction with long-range attention for robust indoor positioning under severe NLoS.
• AI-ML Models for Wireless Positioning Using Channel Information — Indian Patent Application 202541112619 (2025). AI/ML models that extract location-sensitive features directly from channel information to estimate device position.
• Efficient Data Collection and Model Architecture for AI-ML-Based Positioning — Indian Patent Application 202541009543 (2025). Data-collection and model-architecture techniques that streamline training and operation of AI-ML positioning models.
• Method of Improving Accuracy of Positioning a Node in a Cellular Network — U.S. Patent Application US20240414684A1 (2024). Time-orthogonal reference-signal reception across antenna ports plus UE-orientation estimation to sharpen cellular positioning accuracy.
• Low Overhead AI-ML based Positioning Enhancements — Indian Patent Application 202441085443 (2024). AI/ML-assisted positioning that cuts the signaling, measurement, and data-collection overhead of positioning operations.
• AI-ML based Positioning Enhancements — Indian Patent Application 202441074530 (2024). AI/ML mechanisms that enhance positioning procedures using information already available within the wireless network.
• Method of Positioning a Node in a Cellular Network — U.S. Patent Application US20240196361A1 (2024). Joint estimation of ToA, AoA, AoD, and Doppler with antenna/clock/hardware offset calibration and beam prediction.
• Low Complexity Position Estimator for DL-AoD and UL-AoA for B5G and 6G Networks [Runner up: Best Paper Award] — IEEE PIMRC 2024. Low-complexity angle-based position estimator hitting 3GPP Release 16/17 accuracy targets at a fraction of the cost of iterative methods.
• Methods of Determining Position of a Target Node in Side-Link Communication System — U.S. Patent Application US20240089893A1 (2024). Relative AoA/AoD measurements between side-link nodes used to estimate a target node’s position.
• System and Method of Positioning of a Target Node in Side-Link Communication System — U.S. Patent Application US20230292278A1 (2023). PRS-based side-link positioning procedure for V2X, from capability negotiation through position-measurement reporting.
• High Precision Positioning using Multi-cell Massive MIMO system for 5G and beyond — IEEE PIMRC 2021. ESPRIT-based ToA estimation with multilateration across multi-cell massive MIMO, reaching 20 cm accuracy for 90% of UEs in 3GPP indoor-factory scenarios.
• Low Overhead, Low Latency and High Precision Positioning using Wireless Network — Indian Patent Application 202141009458 (2021). Positioning procedures that reduce signaling and measurement overhead while preserving low latency and high precision.