stein@opensky
Another, probably too long, article from me again – hoping to offer some reality checks on 5G. As I have previously commented, there is an ongoing 5G race among operators, among telco suppliers, among device manufacturers – and even among governments and countries, to be the first and best on 5G – and new entrants to the market are also positioning themselves for it.
The bigger question is: To be the first on what? i.e. what is really the 5G everyone is competing for? Technically speaking, 5G has been designed for three main use case areas: 1) Enhanced mobile broadband, 2) Massive IoT – and 3) “Ultra Reliable Low Latency Communication” (i.e. almost instant response communication) – and a number of impressive figures have been quoted, like 100x more capacity than 4G, 10x lower latency (with latencies down to 1 ms), 1000x more devices, etc.
From a real use perspective, futuristic use cases that have been heavily featured include areas such as enhanced entertainment, smart factories, autonomous vehicles, drones, augmented reality, remote surgery, etc – many of which are B2B or B2B2C use cases. In practice, the most obvious use case of “even more broadband” is the one device manufacturers and operators have started with (with a B2C focus). Fixed Wireless Access (FWA) is also a use case that a number of operators have started implementing – as a cost-effective way to provide broadband connectivity as a fixed-line substitution.
As an engineer and an earlier mobile CTO, although I don’t want to be a party pooper, I feel like I need to offer a couple of reality checks (maybe basic, but not necessarily clear to the general public). As a young engineer, I coordinated the tests of candidate (radio) systems for GSM in the mid-80s – and it was an almost religious competition between European telco suppliers at the time about which technology was the best (or most spectrum efficient). Although it was the candidate system from Norwegian Telecom / ELAB that came out best in the tests, the eventual GSM specifications ended up as a compromise between several of the candidate systems, taking the best parts from all, making everyone not too unhappy – and also not giving any company a specific advantage.
So what am I getting at here? As an engineer by training, I believe Shannon’s law and the laws of physics apply. In other words, you can optimize any type of technology, but there is a limit to how much you can squeeze into the physical resources you have, be it time, frequency or space – and energy - and ultimately you will reach the same limit. Other important factors for mobile networks are the distance a signal can travel in different scenarios - and interference between signals. The radio technology used can be FDMA, TDMA, CDMA or whatever – and you can add diversity in terms of MIMO antennas and use this for pointed antenna beams etc – not only saving energy but also limiting interference from other signals.
The 5G standard has been designed with some of the best radio technology that we can imagine today. It has very good spectrum efficiency – but at the end of the day, the practicalities of building a 5G network depends on the actual spectrum you have – and the amount of spectrum you have. Finally, an operator will always consider the costs of building out a 5G network – which depends heavily on the spectrum availability.
Capacity: If you want more capacity in a network, every network planning engineer knows that the easiest way you can do it is to add more spectrum to your base stations (if you have it). If you don’t have any more spectrum, then you will have to add more base stations – which is typically very much more expensive due to extensive Civil Works, need for transport network, site permissions, site leases etc. In an urban area, this is probably justifiable, as the base stations will be filled up in any case. The only downside is the need for more Capex.
Coverage: In less densely populated areas, the situation is different – as the main challenge is coverage. For 5G, using e.g. 700 MHz or 2.100 MHz spectrum will provide similar coverage to previous Gs – and you might even be able to largely make use of existing sites and infrastructure. However, with mmWave spectrum e.g. in the 26 GHz band, the signals don’t travel very far and will struggle to go through walls, vegetation and other obstacles. As a result, the coverage will be very poor in comparison – or you will need many more base stations.
Spectrum availability: To achieve the broadband speeds promised by 5G, you need a very broad band, i.e. you need a minimum amount of spectrum. The only problem is that sufficient spectrum for this is not available in the 700 or 3.500 MHz band (simply because you cannot find e.g. 150 MHz between 700 and 800 MHz) – so mmWave will for very high broadband speeds generally be required (as more spectrum is available in that range). Therefore, to achieve the broadband speeds promised by 5G, it is necessary to use mmWave spectrum – wherever the coverage is planned.
Nationwide coverage: There are probably governments and regulators that dream about nationwide coverage or at least nationwide highway coverage with extreme broadband. Based on the laws of physics, however, operators will naturally build out nationwide coverage using the lower frequencies – but with corresponding limitations in broadband speeds. Although I have not done a concrete business case for it, generally speaking, I believe wide area coverage would therefore mostly be done using 5G spectrum at e.g. 700, 2.100 or 3.500 MHz – or in a combination with 4G. To achieve the promised broadband speeds, mmWave spectrum is required, however, wide area coverage in this band will be extremely expensive. The effective result is that nationwide coverage using mmWave is prohibitive from a cost point of view – and probably even along major highways. It will most likely require some form of public funding – if it is to happen.
Broadband speeds: We have seen various pilots and trials with network suppliers and operators claiming very high bitrates, up to the order of 1 Gbps – and there seems to be a competition around who can demonstrate the highest bitrates. What is not always clear, however, and what people should understand, is that all these tests have been done with mmWave spectrum with no other users or real-life effects present. The following CNN article comments on T-Mobile’s “nationwide” 5G network and its only 20% increase in downlink speed compared to 4G – being a stark contrast to the 100x faster claim. An obvious reason for this is that their coverage is based on 600 MHz spectrum – with the limited bandwidth available in that range.
So far only broadband: The 5G competition seen so far has been on broadband only – and focused on consumers – or, in other words, focused on only the “Enhanced mobile broadband” use case area. The “Massive IoT” and “Ultra Reliable Low Latency Communication (URLLC)” areas are not really covered yet. This is partly because the 3GPP Rel 16 has not yet been available, but also since the “even more broadband” use case is what operators understand. Refer also my earlier articles on “5G? 6G? Hypes, geopolitics and business”, commenting on the main purpose of 5G not being “even more broadband” but digitalization of societies!
While we are waiting for the more future-oriented applications of 5G, let me comment on some of those as well. Let us start with autonomous vehicles. Having travelled and lived abroad for many years, I struggle to see the full extent of autonomous vehicles ever coming along. For example, if you are in the streets of Delhi (or Dhaka), then it is hard to see the relevance of autonomous cars any time soon. With people, children, elephants, cows, bicycles or scooters all over the place, not to mention the road quality, any type of autonomous car would probably either crash into someone or not move at all. In fact, you don’t need to travel very far to see challenges. If you study the traffic in a major junction downtown Paris if the traffic lights don’t work, then chaos emerges instantly.
For connected cars, 6 levels of automation have been defined, in short summarized as L0 “no control”, L1 “hands on”, L2 “hands off”, L3 “eyes off”, L4 “mind off” – and L5 “steering wheel optional”. All cars demoed to date require some form of driver alertness and intervention, with some cars having claimed L3. An example or two of L4 have also been claimed – e.g. along major highway routes in the US, i.e. under very controlled environments.
In my country Norway, autonomous buses in well-defined geographical areas have been tried out and are running in pilots or in commercial use, e.g. in Kongsberg, which also has been a test site for mobile operator Telenor on 5G. Also in Norway, a consortium is doing a commercial pilot with an autonomous bus on a limited area on an island in Trondheim – without any operator involvement. Telenor’s main Nordic competitor Telia recently also announced plans for similar in Stockholm. The focus is on remote- monitored public transport.
Personally, I have not yet seen the role of an operator in autonomous driving (and clearly not in a B2C scenario) – and the automotive industry doesn’t really see a major role for operators with 5G either (except possibly for certain remote monitoring as above – or as a backup if all the other V2X connections fail). For an operator to support autonomous driving it would either require an intense rollout of roadside base stations everywhere (i.e. Capex!) and high use of distributed data processing, storage or caching (close to the vehicles). More likely, however, is that autonomous driving will be implemented with a high degree of direct vehicle-to-vehicle communication and local data processing in the vehicles – and only in fallback solutions mobile network coverage or even satellite coverage will be used.
Communication modes for autonomous vehicles include vehicle-to-vehicle, vehicle-to-infrastructure (road signs, traffic lights etc), vehicle-to-network and vehicle-to-pedestrian communication – and it should be expected that the only part of this for a network operator might be in the vehicle-to-network area. It is still unclear for me what those exact communication needs will be, however. It could potentially include downloading of maps, real-time traffic / routing information, etc.
To be the party pooper again on autonomous vehicles, this might be something that could work for public transport, in well-controlled environments, in case of good roads, with lanes, without snow – and with a certain amount of remote monitoring. It will most likely not work everywhere. The ultimate use of it also requires that all vehicles have autonomous driving features – which will take a long time to get to (not in my lifetime). Autonomous vehicles are definitely cool from a technology standpoint, but its hype seems to exceed the realism. Finally, while mobile operators probably are not best suited as service providers in support of autonomous driving, who else is? The automotive industry? The State? It beats me!
There are good climate and safety ambitions for autonomous vehicles, but there is also a number of safety and liability issues, like: Who is liable if an autonomous car crashes and kills someone? The driver? The car manufacturer? The service provider (whoever that is)? Is there a driver – or is there only a passenger? Does the person in the car need a driver’s license? By the way: Who shall the car hit, if there is a choice? Personally, I drive a L1 Volvo, but I don’t fully trust it. I hold the steering wheel and I am always ready to break – even if the car can actually do it for me.
Massive IoT: Massive IoT in 5G will need to be demonstrated in practice, however, it should be noted that there are already existing technologies like NB-IoT and LTE-M that will cover the needs for IoT for a long time. In many ways, my belief is that IoT could be the most promising use case area going forward, independent of technology. The interesting aspect, in my view, is who will actually be able to make the most out of the opportunity and be able to provide actual services to customers, be it in the consumer IoT or in the B2B area. Having worked for an operator for several decades, I am afraid it might not be the operators though. See also my earlier article on “5G is all around us – but who will eventually benefit?”.
Low latency: Finally, let me comment on URLLC. It has been claimed that 5G provides down to 1 ms of latency - thus allowing almost instant response communication. What should be noted, however, is that this is for the RAN part of the network only. The laws of physics say that for a signal to travel from my city Oslo to Bergen is 1 ms (one way). Correspondingly, it will take 25-30 ms from Oslo to California (one way). What this means in practice is that for proper URLLC use cases, there is a need for “local” processing and storage, i.e. edge computing, depending of the exact real-time requirements.
So what can URLLC be used for? 5G URLLC can fundamentally be used for any kind of application that require low latency and high broadband speeds – and “local” steering / control. Examples could include drones, any kind of industrial applications like smart factories etc – but how it will be provided and who will provide it remains unclear. Questions that come up in such cases include a variety of commercial, legal and regulatory questions – like what kind of SLA will there be if the services are provided by an operator, what liabilities exist, e.g. who is responsible if the network is down, if the whole plant is out of business for a day, if a patient dies, etc. Another question would be how to secure privacy when using camera drones etc - and there are probably many more.
On the other hand, many large corporates are looking for private 5G networks – and are acquiring spectrum for it – in particular in Germany and in some other countries. If these corporates plan to manage these networks themselves, then the liability and SLA issues go away. However, building up a network department for a large corporate is not core business – and will also require skills, competence and Capex. If the corporates outsource to e.g. a telco vendor, then the SLA and liability issues will still remain.
To round off, I have tried to provide some reality checks around some of the 5G hypes. I am sure there are (or will be) good answers to some of the aspects outlined. What remains clear, however, is that 5G will happen – and that there are many great opportunities with it. Who those opportunities will be for, and how the opportunities will be realized, however, remains to be seen. Operators, suppliers and vertical industries all need to prepare, however.
Comentários