Before we delve into the topic, let us first develop a rudimentary understanding of what a 5G network actually is. These are digital cellular networks whose area of service is subdivided into small geographic sections called cells.
5G technology, like 4G, operates on a wide variety of radio spectrum allotments, although it can cover a larger area than existing networks. There are two separate frequency bands in 5G, each of which works in a different way. Sub-6 is the most popular type of 5G; however, there is also mmWave.
Sub-6: Encompasses all 5G operations below 6Ghz. Due to existing 4G LTE networks (which run at lower frequencies), all carriers possess a Sub-6 network in some capacity. The Sub-6 spectrum is critical to the widespread rollout of 5G due to the expansion possibilities without building new cell towers and the ability to travel longer distances and penetrate objects. Essentially, Sub-6 is the component of 5G which allows for better coverage and signal strength.
mmWave: Short for millimeter wave, this is the component of 5G offering with supercharged data transfer rates and low latencies through extremely high-frequency radio waves ranging from 24GHz to 100GHz. The caveat with these ultra-short wavelengths is the limited range and inability to pass simple objects. Hence, the idea of these two frequency bands (Sub-6 and mmWave) is to account for the inefficiencies of each other.
Traditionally, the radio spectrum has not been allocated in the most efficient manner possible. The government divides it into mutually incompatible frequency bands, after which the bands are allocated to various commercial and government agencies for exclusive usage. While the procedure helps services avoid interfering with one another, the owner of a piece of spectrum seldom utilizes it entirely all time. As a result, at any one time, a considerable portion of the allocated frequencies is unusable.
Agencies such as DARPA have sought to solve this spectrum allocation issue through artificial intelligence. The concern behind the initiative was that the increasing application of wireless technologies carries the risk of overcrowded airwaves that our devices require to communicate.
The idea was to create new communication equipment that does not always transmit on the same frequency. The proposed solution was to employ machine-learning techniques to discover the accessible frequencies. They seek to transition from a system controlled by ‘pen and paper’ to one controlled by AI algorithms autonomously.
How AI can help in spectrum allocation:
Using multidimensional correlation across the place, time, context, and state, AI can expose the linkages, dependencies, co-occurrences, and casualties. Thus, reducing alerts to focused, prioritized actions. Some key dimensions to consider are:
AI can correlate events that would otherwise be separated across siloed systems by examining varied data sets across several dimensions. In other words, it has the ability to expose the unseen.
By increasing network quality and providing individualized services, AI is already being utilized to improve customer service and increase consumer experience through chatbots and virtual assistants.
The greatest option for recouping the costs of transitioning networks to 5G is to use AI in network design.
AI efforts are also being applied to improve network performance management.
Managing SLAs, product life cycles, networks, and revenue are some areas where cellular decision-makers want to invest in AI.
Though AI has widespread adoption, 5G can still help bring advancements to the field of AI. For instance, Machine Learning (ML) models require large data volumes to train, and as these models get more complex and powerful, they will need larger datasets.
The low latency and high speed of 5G will allow analysts to swiftly gather, clean, and analyze enormous amounts of data. This will prompt the development of new analytics technologies in the near future.
For example, driverless automobiles were previously limited and a pipedream due to the significant latency supplied by 2G, 3G, and even 4G networks. However, 5G networks will provide minimal latency and improved information processing in real-time. In fact, more broadly, the biggest impact that 5G will have on analytics is real-time data exchange and insights.
Other AI applications such as automation, smart devices, AR, VR, and many others which form the basis of Industry 4.0 will be transformed with the help of 5G.
The 5G architecture is comprised of 3 key service areas:
URLLC offers use cases that need high network dependability (above 99.999%) and extremely low data transfer latency (less than 1 millisecond). As safety requirements demand ultra-reliable connections, data would have to be shared in real-time with minimal delay. Because of the considerable danger involved, autonomous driving, for example, would necessitate such a connection.
Autonomous driving has numerous advantages, ranging from time savings to increased safety due to the elimination of human mistakes. However, all vehicles would need to be connected vehicle-to-vehicle and vehicle-to-infrastructure, such as traffic light systems, emergency services, and road maintenance programs.
Smart factories and Industry 4.0 have comparable requirements, requiring real-time interaction between machinery and robotics. They may also need real-time data from other sensors located throughout the manufacturing facility. Low-latency solutions enable these machine-operated systems to improve manufacturing lines in a safe and effective manner.
Other possible use-cases are remote and augmented reality healthcare, such as remote surgery, smart electricity distribution, and cloud-based gaming and entertainment.
(Read: An Introduction to OpenRAN (ORAN) )
Network slicing (also known as software-defined networking or SDN) will be another important 5G application. In addition to its low latency, it allows telecommunication companies to run several virtual networks on a single physical link. Providers will be able to ‘slice’ the network with 5G, meaning different networks and virtual layers will bring value to the business. Through data monetization, network slicing will enable the creation of new business models.
Each slice functions as its own network, with its own provisioning, security, and service quality needs. As a result, mMTC, which has low security and bandwidth requirements, is isolated from URLLC, which has strong security and reliability requirements. Despite this, all these slices are connected by the same physical network architecture.
As 5G networks adopt AI and thus increase reliance on software, the potential cybersecurity risks related to design flaws (from poor development processes) will begin to matter more. We can already see cases where entities must perform their due diligence to secure their networks. The best example is the ban imposed by many governments on Huawei as a 5G equipment supplier.
Network equipment such as base stations and management functions are becoming more vulnerable to attacks. The dependence of mobile network operators on suppliers means an increase in the possible modes of attack. As a result, suppliers with low-risk profiles will be preferred.
The consequence is that 5G security companies will need to expand to tackle the multidimensional security problem that comes with the next-generation technology. Simply banning a single provider would not be enough.
It will take years to implement fully functioning 5G networks because the connectivity standards have yet to be established, and some aspects of the network have yet to be tested. Some businesses will gradually integrate it into their systems, while other industries, such as Data Analytics, will be quick to embrace 5G. Because it already deals with the challenge of managing petabytes of data that comes with present connectivity, the data analytics business may be the sector where 5 G’s promise will be fully realized. However, with 5 G’s promise of quick and real-time data analyses, the analytics and complex technologies derived from it will bring more potential for improvement.