Open platforms are strategically hugely appealing to operators, promising to stimulate competition and innovation during the difficult migration to virtualized 5G RAN. However, their appeal quickly fades if they cannot be proven to deliver commercial performance that is as good as – or even better than – that of traditional integrated networks.
Rakuten, as the driver-in-chief of the emerging Open RAN ecosystem, has a particularly heavy responsibility to provide a burden of proof. The first operator in the world with a macro-scale commercial Open RAN will be scrutinized – and criticized – beyond all reason by other operators, and by Open RAN critics, eager to understand how this emerging platform really will perform in the challenging applications of 5G.
The early signs were disappointing. First, Rakuten struggled to reach its customer target – despite giving away free access for the first 3m subscribers, it had only reached 2.2m at the end of 2020. There were plenty of critics who questioned the wisdom of deploying such an innovative network if it does not translate into strong revenue and market share growth.
Of course, this ignores the fact that it would have been short-sighted to build a conventional network in a greenfield environment, even if the initial benefits take a while to be seen. And anyway, slow take-up of new services are related to commercial conditions and consumer habits, not to Rakuten’s network architecture. That architecture is designed to reduce cost and to enable agile new services. So far there has been no proof of the latter because Rakuten’s commercial strategy was to launch with a very conventional proposition – free or cheap data to build market share. Later, when it starts to differentiate on the basis of service innovation, the advantages of a cloud network may become clearer.
Second, and more concerningly for the network choices Rakuten had made, were reports that its 5G performance was inferior to that of competitors that, despite progress towards open and virtualized networks, are still burdened with plenty of legacy systems too.
At 5G launch last autumn, Rakuten 5G was lagging behind competitors in average speed and in availability. The latter was often down to its limited coverage in the early phases of roll-out, which mainly apply to parts of the major metro areas of Tokyo, Kanagawa, Saitama, Hokkaido, Osaka and Hyogo. Elsewhere, Rakuten relies on its MVNO agreement with KDDI, but homegrown coverage is improving rapidly. As of the end of January, Rakuten had more than 11,000 base stations covering around 75% of the population, at least with 4G. The aim is to cover 96% of the population by this summer.
But it was the 4G data rates that spooked Open RAN supports. In tests by OpenSignal in September, Rakuten 4G delivered 21.6Mbps on average, compared to 46.9Mbps as an average of the other three MNOs. As Rakuten extends its own network and turns on 5G, things will improve. At launch, Rakuten promised 5G speeds of about 870Mbps at peak, though this did not impress Japanese media and high end users, who have seen gigabit peak speeds from other operators like NTT Docomo.
Rakuten has promised 2Gbps speeds in areas where it switches on millimeter wave spectrum in 2021, and that it will be steadily boosting the peak and average speeds in sub-6 GHz spectrum too with optimization and efficiency techniques.
Last week, Rakuten and its RAN software supplier, Altiostar, announced metrics that supported that claim. Altiostar said recent speed tests showed speeds of 1.77Gbps in millimeter wave spectrum, though it did not say whether these were peak or average rates. This is far higher than the average 5G speeds available to Japanese consumers – in speed tests conducted in the third quarter of 2020 by Ookla, these showed an average download speed of 719.4Mbps was experienced by only the fastest 10% of connections. However, those tests were not confined to millimeter wave spectrum, whose capacity enables very high speeds over short distances.
Citing new tests by OpenSignal, Altiostar said upload speeds had also improved on the Rakuten network, reaching 16.8Mbps – the highest among the four Japanese MNOs – despite Rakuten having “only one-sixth of the spectrum holdings of competing operators in the market”.
Rakuten Mobile’s CTO, Tareq Amin, commented: “The performance and stability of the 5G network shows that a cloud-native framework and webscale architecture can compete with a traditional RAN approach and provide new levels of automation to the network.”
That statement sums up the question that every operator is asking as they evaluate virtualized and open RAN architectures. Many of the question marks apply to any vRAN, not just an open multivendor one like Rakuten’s, but some operators feel a single, highly experienced vendor will address performance challenges more effectively (a view that is heavily encouraged by the incumbent suppliers of course). But others believe the collective weight of testing and innovation that comes from a broader ecosystem, focused specifically on operator requirements, will result in a more robust architecture in time for the mid-2020s, the Open macro RAN witching hour for many advanced MNOs.
Automation will be an important factor in this, and Rakuten says that automation of tasks such as commissioning and integrating new cell sites or identifying network faults will translate into a highly reliable quality of experience, enabled at low operating cost. It says it can provision a new 5G cell site in four minutes and a 4G site in eight minutes.