Deep Learning Based Localization and HO Optimization in 5G NR Networks

Abstract

In the emerging 5G radio networks, beamformingcapable nodes are able to densely cover narrow areas with a high-quality signal. Such systems require high-level handover management system to proactively react to upcoming changes in signal quality, while restricting common issues such as pingponging or fast-shadowing of the signal. The utilization of deep learning in such a system allows for dynamic optimization of the system policies, based directly on the past behavior of the users and their channel responses. Our approach on handover optimization is purely non-deterministic, proving the idea that a self-learning network is able to efficiently manage user mobility in dense network scenario. The proposed network consists of feature extractors and dense layers. The model is trained in two stages, first serves as an initial weight setting in supervised fashion based on 3GPP model. The second stage is an optimization problem to reduce the number of unnecessary handovers while sustaining a high-quality connection. The model is also trained to predict the user location information as the second output. The presented results show that the number of handovers can be significantly reduced without decreasing the throughput of the system. The predicted location of the user has meter-level accuracy.

Deep Learning-Based Cell-Level and Beam-Level Mobility Management System

“…This model is used in LTE networks and serves as the benchmark model in the referred literature [ 2 , 15 , 16 ]. This benchmark model is purely reactive, as it can react to the environment-specific signal changes with a delay of at least 1 measurement period.…”

“…This benchmark model is purely reactive, as it can react to the environment-specific signal changes with a delay of at least 1 measurement period. It is also susceptible to any measurement uncertainties, which cause the model’s performance to quickly degrade as shown earlier in Reference [ 2 ]. Due to this fact, the benchmark 3GPP model’s decision-making may result in numerous ping-pong HOs.…”

“…The basis of the beam-level mobility model proposed in this work is similar to the one proposed in Reference [ 2 ]. Instead of providing mobility over the whole area covered by multiple base stations and their beams, it operates on a single base station for beam-level mobility only.…”

“…The following adjustments were made to optimize the model’s performance and remove redundant information fed to the DL model. The dimensions of the model were reduced when compared to the solution from Reference [ 2 ], as it now serves as 33-target classifier with the reduced number of inputs, and one of the input parameters, namely the serving beam information, was also removed from the model. The number of targets is determined by the number of beams that the BS is able to transmit (32) plus one, indicating that no beam is suitable (lack of BS signal detection by UE).…”

“…The proposed model contains only 2932 trainable parameters. The number of parameters has been reduced compared to the solution from Reference [ 2 ], since the cell-level classifier only chooses between 7 available BS, serving as individual targets for the model, as well as it takes a reduced number of inputs.…”