Advanced computing techniques, often given the summary label of artificial intelligence (AI), and the latest generations of wireless protocols are combining to create new and exciting possibilities. Early generations of wireless connectivity were pursued to enable mobility and eliminate cables. From 1G’s introduction of analog voice mobile telephony, each generation of mobile communications has brought fundamentally new and different innovations. The full impact of these capabilities has typically been unappreciated. That pattern appears to be repeating itself. The potential impact of technological innovations takes time to reveal itself. It tends to surprise all but a few visionary individuals.

In previous wireless generations wireless devices were typically stand-alone generators or receivers of information. Today advanced computing techniques (i.e., AI) are being brought together with wireless to create distributed systems. Data is received from a number of Internet-of-Things (IoT) and other devices. Computation using that data is often done in the cloud. Alternately, in many systems the computational tasks are distributed, with part being done in the data source device but the system analysis being performed at a computational aggregator device, often in the cloud. The results of the computation can then be sent to a variety of sources where it can direct actions or be available to system users.

When developing IMT-2020, the ITU-R (International Telecommunication Union – Radio) focused on three service categories: (3), (4), (5)

• Mobile Broadband
• Massive machine type communications (mMTC)
• Ultra-reliable and low latency communications (URLLC)

Specific services within the triangle of the three general service categories are:

• Gigabytes of user data in a second
• 3D video and ultra-high-definition (UHD) screens
• Cloud based services for work or play
• Augmented reality
• Industrial automation
• Mission critical applications
• Self-driving cars
• Telephony and voice services
• Smart home and buildings
• Smart cities

With the thought of distributed systems in mind, a review of the IMT-2020 use case ‘triangle’ (Figure 3) reveals how central this goal is for IMT-2020 and the 3GPP 5G mobile communication standards that implement it. 

Traditional isolated device services continue to be supported, e.g., voice connections, work and play in the cloud. Most of the triangle is populated by distributed systems. This new generation of wireless communication is focused on enabling distributed systems (e.g., Smart Homes, Buildings and Cities, Augmented Reality, Self-Driving Cars, Intelligent Highway Systems, Industrial Automation).

In evaluating candidate protocols for IMT-2020 the ITU-R used eight measures. IMT measured 5G proposals using eight parameters that are important for those services:

• Traffic capacity
• Peak data rate
• User experienced data rate
• Spectrum efficiency
• Mobility
• Latency
• Connection density
• Network energy efficiency

Several of these use cases reveal a new relationship between wireless and AI. Latency becomes a critical parameter because data must be gathered and analyzed and an action plan developed and transmitted out to the devices that need to act all in a time necessary to serve the system’s function. For some systems the response times are very short. In some systems, such as self-driving cars and medical systems, responses must be fast and reliable. Failure to take a needed action could cost lives.

What is emerging is a new role for wireless and its role in these distributed systems. Wireless must be fast and reliable, but it must also integrate well with other system functions, particularly the AI guiding the system. Some of the chip makers are alert to this new role for wireless as a communication component in a distributed system. NVDIA’s AI-ON-5G platform is one example. As the NVDIA explains:

NVIDIA AI-on-5G is a unified platform that brings together developments in AI and 5G at the edge to accelerate the digital transformation of enterprises across all industries. 5G provides the underlying connectivity for billions of devices, extending the reach of AI algorithms and applications to all connected objects at the edge, enabling new use cases and new markets.

https://www.nvidia.com/en-us/edge-computing/5g/, accessed on September 21, 2021

In these distributed systems the wireless must be tightly coupled to the AI so that the AI is operating from current, complete, and accurate information and is able to direct timely and correct action by the system. Data lost to transmissions that collide with other wireless transmissions have the potential for resulting in poor decisions by the AI engine.

However, beyond the formidable system challenges, there are system-of-systems issues. What are the chances that once self-driving vehicles are developed some of those vehicles are used as ambulances? A self-driving ambulance would free the EMS (Emergency Medical Services) Personnel to focus completely on caring for the patient, letting the ambulance drive itself to the hospital. Perhaps the patient was a victim of an industrial accident at an automated plant. The ambulance would be packed with advanced medical equipment which would be communicating actively with the hospital’s systems so that the hospital is fully informed and prepared when the ambulance arrives. That communication potentially would be directing the care given to this patient while traveling to the hospital. The ambulance starts its journey in the middle of an automated plant with a wide variety of communication controlling the plant operation. It leaves and travels on an intelligent highway system past many smart homes and buildings to arrive at a hospital, almost certainly also in a smart building, packed with advanced medical systems in active use, serving other patients. There are many opportunities for these systems to impact each other in undesirable ways. Just the wireless coexistence challenge is tremendous. Potentially the AI on the ambulance would need to formulate a different communication strategy several times during this journey. Each strategy would be designed to sense the RF environment around it and use bands and protocols that minimize the potential for harmful interference.

What organization will take the lead in addressing these system-of-systems problems? Certainly, the regulators will be involved but they are unlikely to have the cutting-edge expertise required to independently do a thorough analysis or in-depth testing. The most likely solution will be some kind of multi-party cooperation with specialists and resources for a variety of organizations contributing to the effort. The challenges are formidable, but the potential benefits are immense, justifying the necessary effort.

A new role for wireless has been emerging as a communication component enabling distributed AI-powered super-systems. This new role is certainly evident with 5G and will grow in importance with 6G and beyond. With this new role, new parameters get priority and new problems emerge. AI-powered distributed systems hold great promise but bring frightening potential problems as well. The solutions will be systems with tightly coupled AI and wireless with sophisticated spectrum management capabilities that are able to adopt the system communication to the current environment and dynamically change the communication plan as the environment changes.

Bibliography

(1) Leonardo Guevara and Fernando Auat Cheein, “The Role of 5G Technologies: Challenges in Smart Cities and Intelligent Transportation Systems,” Figure 1.

(2) Leonardo Guevara and Fernando Auat Cheein, “The Role of 5G Technologies: Challenges in Smart Cities and Intelligent Transportation Systems,” Figure 2.

(3) ITU-R REC M.2082, September 2015. Available for download at: https://www.itu.int/rec/R-REC-M/en

(4) NGMN (Next Generation Mobile Networks) Alliance, 5G White Paper, February 17, 2015. Available at: https://www.ngmn.org/work-programme/5g-white-paper.html

(5) NGMN, 5G White Paper 2, July 27, 2020. Available at: https://www.ngmn.org/work-programme/5g-white-paper-2.html

(6) ITU-R REC M.2082, September 2015, Figure 2.