It’s no secret that everyone has been talking about AI. It’s the hot topic in every industry. That’s because AI is rapidly changing the way the world does business in every sector. For those of us in the telecommunications space, it’s important to give serious consideration to the network and infrastructure requirements that support current and future AI applications as demand continues to rapidly increase. AI will only be able to grow as far as the network allows it.
Whether for machine learning or real time AI processing, AI needs high-speed, low latency networks to function effectively. We are seeing the use-cases for AI grow in scope and complexity within accelerating time frames, and it’s important that we as an industry invest in the infrastructure needed to keep up with demand.
Some of the most pressing needs include:
– Increased bandwidth that can handle previously unfathomable amounts of data processing.
– Scalable, reliable, elastic and high-capacity telecom connections to keep up with constantly changing demand.
– Ultra-low latency, millisecond level response times.
-IoT connectivity.
-Solutions to energy consumption and sustainability.
1. The Strengths and Weaknesses of Current Telecom Infrastructure
The different network technologies currently operating all have different strengths and weaknesses when it comes to providing connectivity for AI services.
4G is still the workhorse of the telecom world because of how broad the coverage is. 4G can support some of the current use cases such as cloud-based applications, mobile AI assistants, etc. However, because the latencies are typically 30-50 milliseconds, 4G struggles with things like real time AI powered robotics, AR/VR, etc.
5G was rolled out as the solution to technologies requiring ultra-low latency. IoT was hot on everyone’s tongue for a period. Ideally, 5G can achieve as low as 1 millisecond ultra-low latency. 5G is also capable of network slicing. This means that different bandwidth can be allocated to different AI applications. For example, a self-driving car might need an ultra-low latency network slice, while, at the same time, an AI streaming service would need a high bandwidth network slice. 5G is also capable of supporting things like real-time machine learning, AI automation in manufacturing, remote surgery, etc. 5G can support up to 1 million devices per square kilometer, which is fantastic for supporting IoT ecosystems. However, 5G is still being deployed nationwide so there are coverage gaps. Additionally, the mmWave signals have limited range and penetration. 5G also consumes large amounts of power, which limits its sustainability.
Fiber Optic Networks remain the gold standard when it comes to high speed data transmissions, and are necessary support for the 5G network. Fiber is extremely stable and also offers low latency at the range of 1-5 milliseconds. For large scale AI computations and data centers, fiber is absolutely crucial. However, we face large challenges, including a shortage of available long haul fiber, aging networks dating back to the 90s, and the high investment needed to build out new infrastructure.
Satellite Internet is an interesting newer solution, as it provides solutions for rural areas without 5G or fiber coverage. However, the high latency makes the coverage unsuitable for most applications. Exceptions might include things like remote sensing, agricultural use cases, disaster response, etc. The network is still evolving and also has a high cost of expansion.
2. 6G: The Next Generation Network Technology to Support AI
While 5G is a great start to providing faster, more reliable, and low latency connectivity for AI, future investment is still needed as technologies continue to grow and expand. The next frontier–tackling the demands of futuristic technologies like holographic communication and ultra-fast machine learning–is 6G, which is already in research and development stages.
6G will use Terahertz (THz) spectrum for unheard of speeds. Utilizing frequencies above 100GHz, 6G will be able to provide speeds 100x faster than 5G. This will allow for AI use cases that require processing huge datasets in real time like AI powered simulations and virtual reality.
Additionally, latency is projected to drop below 100 microseconds. This would allow for AI to make decisions instantly. This will be key for the next generation of robotics and augmented reality.
Some other interesting future possibilities with 6G include: the ability to integrate AI directly into the network, allowing networks to self-optimize, predict failure, and allocate bandwidth dynamically; the ability to integrate with quantum communications for enhanced security data transmission; using AI to optimize for energy efficiency in the network by predicting demand and switching between high and low power states.
3. The Road Ahead: Public and Private Cooperation
6G is currently projected to be rolled out in 2030, according to Next Gen Alliance, an initiative started by the private sector to develop a roadmap for 6G. Additionally, In March 2019, the Federal Communications Commission (FCC) announced it would open spectrum above 95 GHz to experimental licenses, giving 10 years to conduct these experiments.
It’s of the utmost importance that both the public and private sectors work together along a clear pathway to ensure that 6G and beyond is able to reach its full potential and accomplish the goal of supporting the needs of future AI technologies. The International Telecommunication Union (ITU) released its IMT-2030 Framework, laying a path for unified rollout.
According to Karri Kuoppamaki, Senior Vice President, Advanced and Emerging Technologies, T-Mobile. “The IMT-2030 framework represents a collaborative effort between industry, government, and academia to shape the future of mobile communications. As the industry moves toward 6G, the focus will be on driving innovations that unlock new economic opportunities, enhance connectivity, security, and sustainability while maintaining global leadership in technology development.”
In conclusion, we are facing an exciting journey as we ramp up our networks and infrastructure to be able to support the next generation of AI technologies. We must present a unified front in the development of a 6G network, investment in our long-haul fiber optic networks, and creative approaches to solving high power usage issues. The industry is at the brink of an AI driven boom, and we should plan accordingly.
Author: Jessalyn Gerbholz has over a decade of telecom experience in site development. She was one of the founders of RCG Tower Group, and is currently an industry consultant focusing on network build out in difficult dense urban areas.