A Study on Data-driven Methods for Selection and Evaluation of Beam Subsets in 5G NR

Abstract

5G New Radio is the next generation of mobile networks and it comes with promises of ultra-high speeds, ultra-high reliability and ultra-low latency. This has posed a challenge for the engineers entrusted with the task of finding solutions which could fulfil the specification, and as a result, some promising areas have received increased attention in recent years. Signal processing techniques like directional transmission and reception, also referred to as beamforming, is believed to be a key area of advancement to meet the requirements. The general benefit of applying beamforming is to concentrate the power of a signal in a given direction, which also translates to a smaller area of coverage on the ground surface. As a user traverses the area surrounding a base station, it will thus need to keep an updated image of the radio environment including available beams for transmission and reception. This is done through a frequent and periodic synchronization signal procedure which, partially due to carrier aggregation, intensifies the workload put on the often battery powered user equipment as it continuously updates and stores state variables related to available physical channels. In order to reduce required processing power and cache memory usage in the user equipment, telecommunications engineers have introduced beam subsets which each user may operate on rather than the full set of available beams.

This report investigates purely data-driven and machine learning based alternatives to the current, static, implementation responsible for selecting and evaluating beam subsets with the goal of mitigating the downsides posed by not considering all available channels, as is the case with any such subset strategy. The results show that the current static implementation of the subset selector can be improved in terms of reducing the selected subsets which do not include the widebeam of highest signal strength at a cost of more frequent subset updates, which are of relevance for distributed systems where such configurations take place over the air interface. A probabilistic approach involving a Markov Chain yielded the greatest benefit at the highest subset update cost, while a Machine Learning approach involving a Random Forest Regressor offers a smaller improvement at a much lower cost relative to the Markov method. A subset evaluation technique involving a Multi-Layer Perceptron classifier yielded promising results, being able to detect 76.8\% of subsets not including the strongest widebeam. Another evaluation technique based on a convolutional neural network resulted in accuracies of up to 95\% based on various different image inputs. Furthermore, it was shown via experiments performed on the Markov method that the frequency of selected subsets which did not include the best widebeam decrease exponentially with an increasing subset size.

Popular Abstract

The development of new generations of telecommunications networks is a
continuous process driven by the prospects of improving existing functionality by
for example increasing bit rate and reducing response time. The advancements
of 3G introduced video streaming to your smartphone and 4G took it further by
enabling livestreaming. 5G NR, which stands for 5th Generation New Radio,
is the latest effort by providers of telecommunications equipment to flood the
market with resources needed to both develop existing products and create new
business opportunities in the already gigantic app marketplaces. Customers will
notice the number in the upper left corner shift from a 4 to a 5 and new colorful
icons will be available under the New! tab in their marketplace of choice. The
more observant users might also notice light grey boxes mounted to buildings,
street lights and electrical poles throughout their city and neighborhood. These
are all part of the radio access network (RAN) that enable data transfer between
cellphones and the core network (internet). This article will shine a light on
advancements made within this less colorful, but to the modern smartphone
experience, vastly more fundamental field of research.

The 5G RAN utilizes electromagnetic waves of higher frequency for transmission in
order to increase the number of bits that can be encoded per second.
This part of the spectrum, often referred to as the millimeter wave (mmW)
spectrum, does however suffer from poor propagation distance due to its short
wavelengths. To clarify, the higher up the spectrum we go to reach higher bit
rates, the worse the coverage will become. This physical limitation has forced
engineers to come up with ways to mitigate the negative effects in order to
reap the benefits of the higher frequencies. The aforementioned grey boxes are
a part of this initiative as they are radio access nodes to so-called small cells,
meaning small areas of coverage. Naturally, increasing the density of access
nodes is one way to handle the shorter propagation distance. A more flashy
approach used in macro cells is to concentrate the energy emitted by the access
node in specific directions. This technique is called beamforming and it results
in the signal being divided up into a grid-like structure of coverage areas on
the ground surface, each corresponding to a unique physical channel that can
be used for transmission between the user and radio node. The user equipment and
access node still need to dynamically agree upon which channel to use
for transmission as the user traverses the grid. This is done via a handshake
like procedure in which the cellphone periodically measures the signal quality
of each channel using so-called Tracking Reference Signals (TRS) and reports
back the result to the access node, which can then make a decision and share
it with the user. It just so happens that some macro cells are divided into a
grid so large that today’s cellphones do not have the capacity needed to perform
and report measurements of all available channels. This is solved by assigning
a subset of the grid to each user in the cell, corresponding to the number of
TRS signals the specific cellphone can handle. This summary dives deep into a
study on alternative methods of selection and evaluation of subsets by presenting
data-driven and machine learning based methods that are able to dynamically
select, evaluate and assign subsets to a given user. The study, which refers to
selected subsets which do not include the best channel as beam misses, have
also specified a cost metric, subset switches, which relate to the over-the-air
signalling of subset assignments to the cellphone. The current implementation
of a subset selector which assigns one of four predefined subsets to a cellphone
based on geographical overlap of the subsets was benchmarked. Furthermore,
the study presents two alternative, dynamic subset selection methods. The first
is based on representing the system as a stochastic state machine in the form of
a Markov chain. The second utilizes the a Random Forest Regressor machine
learning model to rank channels based on predicted signal strength. All data
was gathered from a radio environment simulator provided to the authors by
Ericsson. The results showed that near optimal results in terms of average signal
strength could be achieved through the subsets generated by traversing the
Markov chain, which reduced the beam misses by 46%, though at a very high
cost in terms of subset switches. The machine learning approach reduced beam
misses by 12% but at a significantly lower cost than that of the Markov method.
The study also includes testing of the possibility to incorporate evaluation of
subsets into the existing product. They are both based on the training of neural
networks to identify whether a subset is good or bad based on its shape and or
distribution of signal strength within it. With a prediction accuracy of up to
95% for a convolutional neural network approach, the study concludes that an
architecture involving a feedback loop between a selector and an evaluator could
possibly choose subsets very precisely if the increased cost can be tolerated.
In conclusion, thinking outside the light grey 5G box could sometimes lead to
more colorful experiences for the end user.

• author: Ekman, Nic and Skordas, Ilias Theodoros
• supervisor: Björn Landfeldt
• organization: Department of Electrical and Information Technology
• course: EITM01 20221
• year: 2022
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: 5G, NR, telecom, telecommunications, ML, machine learning, algorithms, beam, widebeam, propagation, beamforming, subset, RAN, radio, radio environment, ericsson
• report number: LU/LTH-EIT 2022-888
• language: English
• id: 9098733
• date added to LUP: 2022-09-01 13:57:52
• date last changed: 2022-09-01 15:37:38