[Paper Review] Performance Limits of Single-Anchor mm-Wave Positioning.

Anastasios Kakkavas, Mario H. Castañeda García+2 more|arXiv (Cornell University)|Aug 24, 2018

Indoor and Outdoor Localization Technologies4 citations

TL;DR

This paper derives the Cramér-Rao lower bound (CRLB) for single-anchor mm-Wave positioning using multi-antenna systems, analyzing position, orientation, and velocity estimation under static and dynamic scenarios with and without time-of-transmission (KTT) knowledge. It shows that the Fisher information matrix for position and orientation differs between downlink and uplink only by a scalar factor equal to the ratio of downlink to uplink SNR.

ABSTRACT

The fundamental limits of single-anchor multi-antenna positioning are investigated. Exploiting the structure of the channel at millimeter-wave (mm-Wave) frequencies, the Cramer-Rao lower bound for the position, orientation and velocity estimation error is derived for transmitter and receiver localization under a static and dynamic scenario with or without knowledge of the time of transmission (KTT). After revisiting the relation of the observed multiple input-multiple output (MIMO)-orthogonal frequency division multiplexing (OFDM) channel with the underlying geometry of the channel, we present geometrically intuitive asymptotic expressions for the Fisher information for large bandwidth, large number of antennas and different levels of KTT. Based on our derived results, we show that the Fisher information matrix (FIM) on position and orientation parameters in the downlink (DL) and the uplink (UL) differ only by a scalar, which is equal to the ratio of the receive Signal-to-Noise Ratio (SNR) in the DL and UL.

Motivation & Objective

• To establish fundamental performance limits for single-anchor mm-Wave positioning using multi-antenna systems.
• To analyze the impact of time-of-transmission (KTT) knowledge on estimation accuracy.
• To derive asymptotic expressions for Fisher information under large bandwidth and large antenna array conditions.
• To compare downlink and uplink performance in terms of position and orientation estimation accuracy.

Proposed method

• Derives the Cramér-Rao lower bound (CRLB) for position, orientation, and velocity estimation in mm-Wave systems.
• Models the MIMO-OFDM channel as a function of underlying geometric parameters such as angle of arrival and angle of departure.
• Uses asymptotic analysis for large bandwidth and large number of antennas to derive geometrically intuitive expressions for Fisher information.
• Considers both static and dynamic scenarios with and without KTT.
• Reveals that the Fisher information matrix (FIM) for position and orientation in downlink and uplink differs only by a scalar factor.
• Demonstrates that this scalar is the ratio of downlink to uplink receive SNR.

Experimental results

Research questions

• RQ1 What are the fundamental estimation error limits for single-anchor mm-Wave positioning with multi-antenna arrays?
• RQ2 How does knowledge of the time of transmission affect the achievable positioning accuracy?
• RQ3 How do large bandwidth and large antenna arrays influence the Fisher information for position and orientation?
• RQ4 What is the relationship between downlink and uplink Fisher information matrices for position and orientation estimation?
• RQ5 How does the SNR ratio between downlink and uplink affect the relative performance of positioning in both links?

Key findings

• The Fisher information matrix for position and orientation in downlink and uplink differs only by a scalar factor equal to the ratio of downlink to uplink receive SNR.
• Asymptotic expressions for Fisher information are derived under large bandwidth and large array size, revealing geometric intuition.
• The CRLB for position and orientation estimation is derived for both static and dynamic scenarios with and without KTT.
• Knowledge of the time of transmission (KTT) improves estimation accuracy, particularly in dynamic scenarios.
• The derived CRLB provides a benchmark for evaluating practical mm-Wave positioning algorithms.
• The results show that downlink and uplink performance are symmetric up to an SNR-dependent scalar, enabling performance prediction across links.