5G bans point to bigger changes in design

A political decision has finally put paid to Huawei’s ambitions to sell its gear to UK cellular operators, although it might take some time for the ban to close down the supply pipeline completely. 

A number of hardline Conservative MPs are unhappy with the seven-year transition period for UK operators to remove Huawei gear from their 5G networks. But with no compulsion to rip out gear for older generations, it is easy to conclude from the UK government's action that there is no clear and present danger that requires an immediate fix. US president Donald Trump’s willingness to take credit for the decision points just as strongly to the move being more one of using security as a convenient rationale for what, ultimately, is being driven by trade strategy.

One argument for the sudden reversal in UK policy was that Huawei, having been banned from using US silicon, would have to use supposedly less verifiable chips made in the Far East. However, as most advanced production for these devices is in Taiwan and South Korea, it is not the strongest argument. One would expect both types of silicon to go through the same kinds of searches for backdoors and security flaws no matter where they are made, given that Trojan circuits can be inserted at multiple points in the supply chain – many of which would not be under direct control of a single government. Security researchers have shown time and again that 'fuzzing'-type attacks are relentlessly effective in revealing backdoors.

With Huawei in particular, you also have to consider where most of that hardware was going: into the front-end radio units rather than the core network. Exfiltrating data from even compromised units seems a tall order. And even the prospect of radio units obeying remotely issued kill commands in the event of a conflict grows more remote as the focus on security in the core network increases. Other vendors’ equipment might be vulnerable to denial-of-service or kill commands for different reasons. 

It seems strange, even in 2020, that mobile communications should become a major battlefield in a trade and technology turf war. But it’s far from being the first time. Cellular technology has repeatedly been an exercise in countries aiming to gain competitive advantage in a key market: usually making it more difficult to communicate. 

It is easy to cite ETSI’s approach to standardising GSM as an example of how state- or region-level intervention can work and in turn promote local players. Before 2G, Nokia’s main investments were in wood and rubber. Today, it remains one of the few companies left that can supply a complete suite of cellular infrastructure equipment – though this may be about to change, courtesy of the latest geopolitical shenanigans. 

While GSM surged ahead around the world, the US persevered with its home-grown CDMA technology not just for 2G but for 3G in what was often a fractious standardisation process as companies based there realised how much ground they stood to lose. In the event, GSM won the economic war for 2G even in North America. Japan, which had bet heavily on its own local 2G protocol, also lost out.

Two decades ago, long before Xi Jinping took charge and while the country was considered to be opening up its economy, China also contributed to the balkanisation of 3G. Its engineers worked with Siemens to develop its own variant called TD-SCDMA that China could control rather than falling in with either the not-so-different European flavour or the CDMA2000 option from the Americans. By the time 4G rolled around, there was a much greater focus on global cooperation in the spirit of the times as 3GPP emerged as the key clearing house for patents and protocol proposals. Almost a decade ago, China Mobile, which had the key licence for TD-SCDMA, was pulling back on investment in the standard. It had become the Betamax to WCDMA’s VHS. It looked as though everyone was happy to avoid splintering a standard to try to protect local manufacturers. 

For 5G, the landscape has changed completely, though old-fashioned ideas on competition persist. This time, as with 4G, there is no realistic option for splintering standards. To a large extent, 5G is more about bringing a lot of different radio standards under one roof and using real-time computation to work out which one each user gets. Their devices may switch between Wi-Fi, 5G New Radio and 5G millimetre-wave without any obvious glitches: the software in the infrastructure will sort it out.

Instead, sanctions mean two of the top-five suppliers have been knocked out of the running in the West –though they seem likely to gain market share in countries where China can exercise its influence – and operators have to find other options that are compatible but not designed in Shenzhen. How practical that will be is a matter of debate. 

The embargo on Chinese hardware for 5G seems at first to be a boost for the likes of Ericsson, Nokia and Samsung. But if it seems odd that the US would be putting in place a strategy that promotes a bunch of non-American suppliers, the reality it that it does not expect things to pan out that way. A takeaway from the development of 2G, 4G and then 5G is the way in which North American suppliers have disappeared. But now they have a way back in thanks partly to a likely slowdown in rollouts and the growth of cloud-computing technologies. 

A number of startups and integrators such as IBM are putting their weight behind the Open-RAN architecture. This is a move analogous to the replacement of big-iron computer servers by commodity PC hardware, and follows on from a similar trend in wireline infrastructure known as software-defined networking (SDN). In this architecture, you push most of the digital processing needed into general-purpose processors of the kind made by Intel, Marvell, Qualcomm and others. This is easier said than done.

It’s worth noting that in SDN, though you can get a rack of Intel x86s to do the job, many operators are using more specialised processors where they can to get better power efficiency and lower cost. However, because Open-RAN is based on multi-vendor standards, it is in principle much easier for operators to swap out radio units and baseband controllers than it is with the equipment that infrastructure vendors have provided traditionally. This move does not leave the incumbents high and dry. Nokia has become an enthusiastic backer of Open-RAN in recent months. But it is a move that provides a way for US suppliers to make the most of their experience in cloud computing to encroach on a market they could not previously sell into easily. They should probably keep a keen eye on security, though.