Boosting performance in 5G networks

Some months ago, we wrote a post on the advantages of using Quside’s RPU technology to improve MIMO decoding for 5G technologies. Since then, we have had the opportunity to discuss these advantages with several telecommunications operators and radio access technology providers, learn where the technology fits better in the 5G environment, and advance in revealing the business benefits of these innovations for Telco operators.

Challenges

With the pervasive deployment of 5G technology and with 6G technologies already on the horizon, infrastructure operators face several challenges, the most prominent being:

• Triple-digit growth in 5G connections. This growth trajectory aligns with the increasing demand for mobile data. Over 6 billion connections are expected by 2027.
• Increased number of cells due to lower range. The mmWave frequencies used by 5G cover only short distances, demanding an ultra-dense grid of small cells to provide a similar coverage area as previous technologies.
• Higher OPEX costs and initial CAPEX investment. Including ground lease and expenses associated with configuring, testing, managing, and maintaining the network.
• Performance limitations in MIMOs. High bit error rates (BER) compromise network efficiency and reliability, resulting in low Quality of Service (QoS) metrics and a heavily degraded service.

The importance of equalizers

In the 5G stack, many interconnected pieces play their role in guaranteeing the performance and quality expected from the wireless network. Among those pieces, the equalizer is one of the most critical, overseeing converting the analogic signals captured by the antenna into the digital domain. The equalizer role is not exempt from challenges, though: it must simultaneously distinguish between the signals of all the users communicating with the antenna while at the same time removing all the noise that distorts them. In this challenging context, any improvement in the equalizer is extraordinarily relevant in optimizing the network, as it may result in several advantages such as:

• Enhancing signal clarity: Equalizers play a pivotal role in telco systems as critical parts in recovering the signal from the emitter. The better the equalizer performance, the better the quality of the recovered signals.
• Adaptive equalization for dynamic environments: Implementing adaptive equalization techniques allows telco systems to adjust to varying signal conditions in real time, ensuring consistent performance.
• Impact on infrastructure needs: Improved equalization reduces the necessity for extensive physical infrastructure, leading to cost savings and more efficient network deployments.
• Future-ready networks: Advanced equalization paves the way for the evolving demands of 5G, 6G, and beyond, preparing networks for higher data throughput and lower latency requirements.

Using stochastic algorithms to boost performance

While equalizers are major in network performance, existing equalizer methods have several limitations. They typically rely on matrix inversion techniques that make the equalization computationally intensive and complex and, at the same time, numerically unstable. That makes current equalizers limited in performance and computationally inefficient.

Taking advantage of the randomness processing power of Quside’s RPU, Quside’s MIMO Equalizer can show reductions of up to 40% in the infrastructure required to support the 5G service while providing the same quality of service. How do we do it?

• By exploiting Quside’s hardware-accelerated stochastic solver
• Using stochastic methods

Accelerating MIMO equalization for 5G networks with Quside’s Stochastic Equalizer

In our previous post, we show the improvement in the Bit Error Rate (BER) performance that Quside’s Stochastic Equalizer provides compared to commonly used deterministic algorithms, such as Zero Forcing (ZF).

We saw at that time that ZF presents limited performance and inefficiencies that bound the number of devices the system can manage to keep BER and latency under control.

MIMO equalization on Quside’s RPU with Quside’s Stochastic Equalizer obtains a reduced BER with an increasing number of managed devices independently

of the noise, keeping latency under 5 ms, while boosting transmission reliability.

That traduces at the end of the day to an increased number of devices or transmission speed. As an alternative, we can reduce up to 40% the number of antennas required to keep almost the same coverage. In a recent work, we simulated with DeepMIMO’s Street Crossing Scenarioⁱⁱ the performance and infrastructure requirements provided in a typical urban area layout with 18 antennas distributed across a few buildings and a city crossing.

In one case, using traditional ZF equalizers, the level of service provided with 18 antennas gets severely impacted by the reduction in antennas (either due to failure or removal). This impact is the effect of the increases in Bit Error Rates (BER) and the equalizer’s computational underperformance.

Using Quside’s MIMO Equalizers with Quside’s RPU, the level of service provided not only increases but is maintained even while reducing the number of antennas.  Our stochastic algorithms can remarkably compensate for BER shortcomings even in a low-latency environment.

The summarized results found that, even in the extreme cases of shutting down half of the antennae in the same DeepMIMO scenario, Quside’s Stochastic Equalizer algorithm proves better QoS metrics than its deterministic counterpart. For ZF algorithms, halving the number of antennae reduces the coverage to 72.58%, which means that 27.42% of the users remain uncovered. In contrast, when using Quside’s Stochastic Equalizer, the coverage reaches 88.89%, with only 11.11% of uncovered users. The results show an improvement of 22.47% in coverage in case of a reduction of 50% in antennas.

In conclusion, we demonstrated that introducing modern stochastic algorithms, combined with Quside’s RPU accelerator in the key equalizer component of the Radio Access infrastructure, provides huge advantages in optimizing 5G deployments.