Performance of Offloading Strategies in Collocated Deployments of Millimeter Wave NR-U Technology

TL;DR

This paper evaluates offloading strategies for millimeter wave NR-U deployments coexisting with WiGig, revealing that simple offloading policies perform nearly as well as more complex ones in terms of session loss probability.

Contribution

It introduces a stochastic geometry and queuing theory-based framework to analyze NR-U/WiGig coexistence and compares various offloading strategies in mmWave environments.

Findings

Baseline offloading strategy yields best performance.

Heavy session offloading causes minimal performance loss.

Offloading smaller sessions results in worst performance.

Abstract

5G New Radio (NR) technology operating in millimeter wave (mmWave) band is expected to be utilized in areas with high and fluctuating traffic demands such as city squares, shopping malls, etc. The latter may result in quality of service (QoS) violations. To deal with this challenge, 3GPP has recently proposed NR unlicensed (NR-U) technology that may utilize 60 GHz frequency band. In this paper, we investigate the deployment of NR-U base stations (BS) simultaneously operating in licensed and unlicensed mmWave bands in presence of competing WiGig traffic, where NR-U users may use unlicensed band as long as session rate requirements are met. To this aim, we utilize the tools of stochastic geometry, Markov chains, and queuing systems with random resource requirements to simultaneously capture NR-U/WiGig coexistence mechanism and session service dynamics in the presence of mmWave-specific channel impairments. We then proceed comparing performance of different offloading strategies by utilizing the eventual session loss probability as the main metric of interest. Our results show non-trivial behaviour of the collision probability in the unlicensed band as compared to lower frequency systems. The baseline strategy, where a session is offloaded onto unlicensed band only when there are no resources available in the licensed one, leads to the best performance. The offloading strategy, where sessions with heavier-than-average requirements are immediately directed onto unlicensed band results in just $2-5\%$ performance loss. The worst performance is observed when sessions with smaller-than-average requirements are offloaded onto unlicensed band.