5G technology is set to define the cellular networks for the next decade through 2030 at least. The definition of the 5G standard is intended to future-proof the technology, hopefully without over-engineering either the standards or the technology. Even though 4G is the most successful and rapidly adopted cellular technology ever, the 5G standards body learned from that experience and not repeat those mistakes. The use case (or killer app) for 4G was the USB dongle, but then the smartphone and mobile video burst onto the scene. 4G did not anticipate the heavy video demand or else the industry would not have insisted on balance uplink and downlink spectrum. A 3GPP 5G goal is to define the technology good for 10 years of innovation, and since they can’t anticipate all future use cases, they put in specify the enabling technology foundations for unanticipated 5G use cases. Hence, we see all sorts of fanciful visions of 5G that quite frankly appear hallucinogenic, but by 2025, might well be required table stakes.
A foundational pillar of 5G is ultra-reliable low latency communications (URLLC), and this will enable all manner of use cases including one that comes frequently to mind, the autonomous, self-driving vehicle. The 5G hype asserts as self-evident that autonomous, self-driving vehicles will need 5G and URLLC. Such presumption is often unsubstantiated. A hint of this comes analyst from the use of “certainly” or “surely” or “clearly” when addressing the technology, which points to the unsubstantiated assumptions. These may that just sound right to the analyst, or maybe the analyst is just making it up.
Not Yet Ready for Self-Driving Cars
The self-driving vehicle community is not waiting for 5G and progressing quite nicely with its own initiatives. Shown below are the two development vehicles seen in and around the Phoenix area from Uber and Waymo. GM is also making its presence known.
Note that safe driving practices were used and these photos obtained while stopped in traffic – you can see the traffic signals above each of the vehicles, one having just turned green. The Uber vehicles seem ubiquitous on the streets around the local university and both vendors have a driver in vehicle, so it is not obvious when autonomous driving is active. The swarms of students exiting their classrooms and onto the streets suggests that might a condition be best suited for a human in control loop. The Waymo vehicle on the right is seen waymo less often than the Uber vehicle, at least where and when I drive.
Is 5G URLLC necessary for self-driving vehicles?
Low Latency control of the self-driving car depends on the events at hand, and the low latency command and control loop must reflect immediacy of that event. The real-time, life-critical issues of stop, go, turn, accelerate, etc. must reside entirely on the vehicle. There cannot be a dependency on any outside network system to control the fundamental responses of the vehicle. Very few insurance companies or corporate boards will sign up for this once they understand the implications of a network outage. Even an Ultra-Reliable network will still have those moments of failure and it doesn’t take much imagination to grasp the tremendous legal risks. This self-driving car of the future must be mini-data center on wheels, and capable of autonomous self-driving without 5G – just like they are learning to do now without 4G. Self-driving vehicles must have autonomous self-driving capability regardless of the network certain conditions.
The 5G Value: Edge Computing & Sensor Networks
There are other events that will come into play as standards are put in place and 5G infrastructure reaches a coverage and density comparable to today’s 4G networks. These will have URLLC needs and the vehicle must fall back to a completely autonomous mode when the 5G network is not available. This 5G enablement is cooperative autonomous driving, such as platooning, where traffic conditions are observed via sensor networks and predictive analysis are put into action. Bringing this about will require close cooperation among auto and insurance companies, regulators, and different entities of local, state, and federal agencies to name a few.
It is inefficient and impractical for each vehicle to query hundreds of sensors with thousands of queries and data handshakes for events such as at intersections as the basis for coordinated driving. Instead, intermediation and aggregation of sensor data and high-level analytics will establish macro-level commands to deliver cooperative driving. This is where edge computing comes to the front, providing the low latency and localized computing. The common use of the highway and street system also implies a single system for command and control of cooperative driving. Otherwise, chaos would reign if there were four national operators deploying 4 mobile edge computing facilities and four separate 5G networks and hoping that coordinated autonomous deriving results. This is a scenario where a natural monopoly makes sense, as is the case for a local utility companies.
4G networks can serve applications that do not have stringent local timing requirements, i.e. millisecond latency when 5G networks are not available. The 4G networks can still provide high-level traffic routing and guidance and there is less need for a natural monopoly or host-neutral-system.
Edge computing is not a given, as it must still pass the CFO test, which is to provide a better ROI than other investments or what was done previously. While edge computing is feasible, it must gain the mobile operators’ interest and generate a business case that attracts the investment. Monetizing edge computing could happen many ways and represents opportunities for innovation as does the 5G technology itself. One can imagine direct subscriptions, insurance sponsored plans, special auto manufacturer bundles (comes with SkyNet!) or even user taxation which could be the electronic equivalent of HOV stickers.
5G will efficiently deliver information and entertainment to meet the needs and wants of autonomous vehicle passengers. With the exponential growth of mobile broadband video, 5G mmWave can cost-effectively handle the traffic load and artificial intelligence-driven networks will improve the user experience. For example, a vehicle leaving 5G coverage or moving into a high congestion area could receive a boost in download capacity before the end of coverage. Monetizing infotainment is easy to imagine, and is understood by the transportation and entertainment industries. 3G based mobile infotainment was neither practical nor affordable for the mass-market, and with 4G, it is feasible for the high-end consumer. As 5G mmWave becomes pervasive (at least in urban areas) and penetrates the mass-market, mobile infotainment will become table stakes in the era of autonomous self-driving vehicles.
Finally, “imagination is more important than knowledge” and the intersection of 5G mobile broadband, autonomous vehicles, and entertainment will open more business innovation that we can imagine, and this is precisely the point of the 5G standards bodies. Imagine this: autonomous RoboCars also operating as mobile communications and data centers equipped with 5G fat pipes. Here mobile network means the network that is mobile, with the RoboCars forming a mobile mesh network on the move and adapting to both physical automotive traffic and the mobile broadband traffic. We have UberEATS (here) in some markets, so it’s not a stretch to imagine an UberHotSpot, where 5G connections to the RoboCars provide backhaul, and traffic is then distributed locally via licensed or unlicensed spectrum. If Alphabet Loon deliver cellular service with stratospheric balloons, then Uber (or someone else) could manage this and route the RoboCars to relieve hotspot congestion. The destinations for high ride-hailing and high mobile data traffic are naturally complementary.
Finding Value in the 5G Machine Economy
This imagination drives the hype cycle of new technology, and 5G is no different. The 3GPP 5G standards effort is correct to focus on the capabilities needed to make the technology long lasting and future-proof, though that leads to the excessive hype that now accompanies 5G developments. The development of advanced technologies like 5G should cause vendors, operators, and enterprise to consider what their roles will be and how their business models must change as the global economy moves from the Internet Age into the Machine Economy. Understanding the industry value chains and how they map into the future is the study of Machinomics research at SAR Insight. Follow us as we follow the money into the Machine Economy.