The advantages of network sharing seem glaringly obvious in a world where the mismatch between mobile data demand and mobile data ARPU is rushing MNOs’ profits. The need to reduce the cost of delivering those rising tides of data is urgent, but many operators are ready to discuss almost any tactic – WiFi offload, automation, outsourcing, even an early move to more spectrally efficient 5G radios, or an outright merger – rather than consider sharing the RAN load with others. So will that reluctance to lose control and ‘enable the competition’ end up weakening the 5G business case, delaying deployments and letting non-MNOs, with a more open approach to infrastructure, into the market?
Even in the rising number of markets where regulators permit sharing of active network equipment, the level of RAN sharing remains low (by contrast, there is significant willingness to share passive infrastructure such as towers and transport links). For every operator which has pushed ahead with RAN sharing, such as the UK’s big four, there are several others which reject the notion. This month, AT&T was the latest to argue against the idea that building three or four 5G networks per country would be the only way to get 5G rolled out.
The US operator’s chief strategy officer John Donovan told the AT&T Developer Summit, held on the sidelines of the Consumer Electronics Show, that “the potential to do a big quality degradation is too high” to justify RAN sharing, especially when 5G is being built out in high frequency spectrum, where signals have short range and there is high potential for interference. He argued that the risk of an impact on quality of service would not be justified by “an economic benefit that’s unknown and not that big”. He claimed that there is little evidence that existing RAN sharing arrangements, such as those in some European countries, have resulted in significant cost savings or market gains.
Donovan was referring mainly to a specific use case in emerging millimeter wave spectrum, but his more general comments about quality risks and unproven cost savings go to the heart of many operators’ resistance to sharing. Yet, as many start to pursue new network architecture strategies, such as densifying with large numbers of small cells, and virtualizing baseband processing in a Cloud-RAN, the case for sharing the load is likely to become more pressing.
In the former case, for instance, calculations by Rethink’s RAN Research service concluded that, if the percentage of small cell networks which are multi-operator or neutral host stayed at its current level until 2020, the installed base of non-residential small cells could be more than 4 million units, or 32%, smaller than indicated in current forecasts which assume a rising level of multi-operator support. Even though shared networks involve fewer cells than multiple single-operator ones, nevertheless the total deployments would be lower because so many networks would not be commercially viable without sharing or neutral host (where an external party builds and runs the network and the MNOs act as tenants).
The sheer number of sites, backhaul links and power sources which would be required to support multiple dense networks in the same place would mitigate against this and make a clear argument for a single, shared network, whether built by one of the MNOs, a joint venture between several MNOs, or a neutral host.
Where none of those shared approaches can be agreed, there will often be no network at all. Densification will be reduced or abandoned in favor of other methods of increasing mobile capacity, such as WiFi offload or new spectrum; or some areas, particularly rural and in-building, will go without decent coverage and capacity.
Risks like these lurked behind former FCC chairman Tom Wheeler’s words late last year, when he told a conference that there are just over 200,000 cell towers in the US, but there may be “millions of small cell sites in the 5G future”, making it essential that the industry “think creatively about smart solutions to the deployment of the antennas necessary for 5G to benefit the public”. He added: “To be clear, I’m not endorsing shared infrastructure in every and all circumstances, and certainly not opening a door to consolidation. But I am saying that if we’re talking about thousands of antennas in a city, and you’ve got four carriers, and we are serious about leading the world in 5G deployment in our very large and spread-out country, we ought to explore creative options on how best to build that infrastructure.”
Despite the obvious sense of such words, MNOs like AT&T remain protective of their control over their networks and how they deploy and monetize them. In a survey of over 40 mobile operators which Rethink conducted in mid-2016, the reasons for MNOs’ suspicion of RAN sharing emerged. Although the survey related mainly to small cells, many of the issues apply to the macro network too.
The primary issue is that MNOs see too little incentive to participate in building a network which will, in the view of many, just ‘enable their competitors’. While most MNOs have embraced the MVNO model to increase and diversify their revenue streams, few go as far as to share their networks or spectrum with direct MNO rivals beyond roaming deals, unless mandated to do so by government.
As Nick Johnson, CTO of small cell vendor ip.access, put it, multi-operator networking “essentially opens the door to a free-for-all in terms of resources. There’s no commercial benefit to the first mover. All the first mover does is open the door to their competitors. So there are no first movers.”
The Rethink survey found that the need to take the upfront capex burden, and uncertainty about the ROI, are the most important barriers deterring MNOs from sharing networks. Other factors which were commonly cited included fear of enabling rivals (a top three concern for over half of respondents); concerns about differentiation from those rivals (38%); and issues of network congestion (32%).
They also have qualms about being tenants on a network hosted by another, even a neutral third party. Here, the concerns are also centered on uncertain monetization models (placed in the top three by 51% of MNOs), as well as the risk of having limited control over quality of experience (45%), or over differentiation (38%) and lack of control over QoE (32%).
When it comes to Cloud-RAN, there will be even more fundamental decisions to make. The more pure – or extreme – interpretations of C-RAN will see very stripped-back cell site equipment, so it may become more practical for each operator to have their own, though there will still be site issues when those cells are small. The bigger decision will be whether operators will own and run their own cloud infrastructure; share those data centers with others; or have their virtual RANs hosted by third party cloud giants.
For now, though, most operators are concentrating on their current architectures and the pros and cons of sharing a 4G macro network. One of the factors which will drive sharing, of course, will be government pressure or even mandates. Some countries have sought to speed up LTE availability by backing a wholesale, shared network, or forcing existing operators to band together in rolling out 4G.
These schemes have had mixed results. A plan for a neutral host LTE network in Russia, to be funded and shared by the three major MNOs and the wireline incumbent Rostelecom, fell apart amid wrangling between the stakeholders – finally one of the MNOs acquired the neutral host and all the operators built their own 4G systems. Shared or wholesale network plans in Kenya and South Africa ran up against political opposition and unwillingness by MNOs to surrender control over their timescales, investment levels and QoS.
One of these government-mandated schemes which is coming to fruition is in Mexico, where regulator Promtel last week signed a public-private partnership deal with Altan Redes, which will roll out LTE infrastructure, in dedicated 700 MHz. The network is expected to cost around $7bn and Altan has promised to cover 92.2% of the population, more than the 85% minimum target specified in the tender. The network must begin commercial operations no later than March 31 2018, and will cover 30% of the population at launch.
The agreement was delayed and mired in controversy, but now it is signed the wholesale national network could open the market to new competitors to operate alongside the existing 4G providers, notably America Movil, AT&T and Telefonica.
Altan Redes, a consortium led by funds owned by Morgan Stanley, the World Bank and Spanish businessman Eugenio Galdon, won its 20-year licence to deploy and operate the network, called Red Compartida, in November, beating the only other bidder, Rivada Networks, which was disqualified for allegedly failing to provide necessary financial guarantees. Rivada is still disputing that decision and recently argued that, since the Chinese government indirectly holds 23.6% of Altan’s equity, the PPP is in violation of the rules which bar ownership of Red Compartida by foreign governments.
Despite these wrangles, there are hopes that Red Compartida will introduce new competition and services to Latin America’s second largest telecoms market, where innovation has been limited by the stranglehold of incumbent America Movil. The dominance of that company has been attacked in the past two years by a new regulator, which has forced the firm to divest some assets, and by the entry of AT&T. And now the wholesale network could support new service providers such as cablecos, as well as MNOs, or could enable AT&T – which acquired two smaller mobile carriers, Iusacell and Nextel – to extend its capacity (its purchase of DirecTV also brought some Mexican activities).
However, the biggest challenge will be the difficulty of competing with established incumbents – especially with a network which will not even start service for another 16 months – if the MNOs do not choose to use the wholesale system alongside their own. If cablecos and other providers get sufficiently interested in multiplay services, there should be rising custom for a wholesale network, but in many areas it is tough to make a business case for investing billions of dollars if the MNOs do not become clients in their own right. The length of time for new entrants to generate significant revenue, and their small market share, are significant barriers to breaking down the role of the established majors with their long history of investment in tightly controlled RANs. And in most cases they are reluctant to weaken that control, even in the interests of slashing their capex and opex budgets.
The shared network may also be too little too late, and the tortured process of awarding the contract has highlighted the barriers to the government-mandated wholesale or sharing approach. The government often gets involved in a bid to bridge the digital divide, especially in rural areas. In these areas, the case for sharing seem even more obvious, since it is hard to justify the cost of building a fully-fledged mobile broadband network for a sparsely populated, hard-to-reach neighbourhood. Operators will often voluntarily address rural shortfall by agreeing to passive infrastructure sharing – there were 64 such agreements in place in Asia-Pacific in mid-2016, for instance – but tend to remain hostile to active RAN sharing, even thought that would deliver even greater capex and opex savings.
This is seen in Australia, despite that country’s huge interior representing one of the biggest challenges for affordable mobile broadband coverage. In 2011, incumbent Telstra ended its RAN sharing agreement with rival VHA (Vodafone Hutchison), with chief scientist Hugh Bradlow dismissing sharing as a “race to the bottom”, saying: “What happens with network sharing is that nobody is incentivized to invest in the networks but the marketing departments promise as much bandwidth as possible.” Telstra has been notable, in recent years, for differentiating itself by staying at the cutting edge of network technology, deploying successive 4G advances at an early stage and promising to be in the 5G vanguard too.
Last October, Telstra was attacking network sharing from another angle, fighting against government proposals to force it to open up its network to rivals, especially in rural areas. This would be for compulsory roaming rather than full RAN sharing, but this approach to universal coverage is just as mistrusted by many MNOs, as seen by the furious backlash of the UK’s four operators against government calls for mandatory rural roaming.
Telstra’s network reaches 98% of the Australian population and it negotiates third party access on a commercial, negotiated basis. Two previous enquiries by the Australian competition watchdog had ruled that there was no need to regulate the charges, but rivals like Vodafone VHA have been claiming that the fees are too high, while rural communities and farmers want to have more than one choice of service provider.
These demands have resulted in a third enquiry by the competition regulator, the ACCC. It is considering ‘declaring’ wholesale domestic mobile roaming services (making them compulsory on request at prices and terms set by the ACCC). Among 13 existing declared services in Australia are wholesale ADSL and superfast broadband access.
Telstra’s chairman, John Mullen, hit back at an October annual general meeting, saying: “Why would anyone invest in maintaining or upgrading their regional networks when they can hitch a ride on someone else’s network and there is no longer any competitive differentiation from greater network coverage?”
The operator believes that its investment in covering most of the population in such a difficult terrain is key to its competitive edge. Tony Warren, head of corporate affairs, told journalists that Telstra had invested huge amounts in that, and “at the moment if I am a customer who doesn’t value coverage, but values price, I go with Vodafone. Australians value coverage and that’s why they come to us.”
He added: “We have invested in a whole lot of towers in the bush that are not economic in their own right. But because they have extended our coverage claim it has made sense for us to upgrade and extend. We made a choice to be the premium supplier with the best coverage and the best network. What Vodafone is trying to do is have that removed by regulation.”
And CEO Andrew Penn added: “The principal advocate for mobile roaming is a foreign company [Vodafone] that has chosen not to invest to the extent that Telstra has. A foreign company that is very capable of investing and a foreign company that has argued against roaming in other markets where it suits it do so.”
The second largest MNO in Australia, Optus, is also arguing that compulsory roaming and regulated pricing could pose a risk to investment in rural areas.
But consumer groups, and smaller operators like TPG, point out that rural dwellers are not actively choosing Telstra with its higher prices – they have no choice. Telstra’s prices are, on average, 15% higher than its competitors’.
Telstra opponents also point to the subsidies the incumbent has received for extending its rural network. For instance, it was awarded $40m to add 113 towers along regional highways in Western Australia in 2012, and $95m to build 430 new towers under the federal Mobile Black Spot program. It did add even larger sums of its own money (Telstra spent A$1.3bn on its network in 2015), but nevertheless, critics say the subsidies should have come with requirement to share the sites with others.
“Declaring domestic roaming would remove all the incentives for current record levels of investment in mobile infrastructure in regional areas,” Optus’ VP of corporate and regulatory affairs, David Epstein, said, adding that the SingTel-owned firm would not have spent its 2015 sum of A$1.7bn on mobile infrastructure “if it was going to be subjected to domestic roaming declaration”.
An ACCC preliminary ruling is expected early this year.
While these rural roaming disputes seem a far cry from discussions of sharing dense pre-5G HetNets, the underlying themes are the same – the operator’s ability to differentiate through its coverage and the way it enhances and manages its network, versus the consumer’s ability to choose between different service providers and tariffs. That basic dilemma is there from 2G to 5G, and from urban to enterprise to rural networks. And as the mobile network supports more and more use cases, including specialized mobile-first activities in vertical markets and the Internet of Things, the choices will only get more complex.
AT&T’s Donovan argued that operators with very different commercial strategies – one focusing on simple daily communications, for instance, and another prioritizing critical medical systems—could not be effectively supported by the same network.
The answer to that, however, is not to turn away from sharing and wholesaling, but to embrace it. Flexible, on-demand wholesale platform supporting large numbers of specialized service providers is the way forward to ensure return on 5G investments. The density of 5G in some areas will absolutely require sharing – a shared or neutral host approach will be the only one possible in locations where hundreds of cells are needed per square kilometer. Yet there will, as Donovan said, be challenges when many viable 5G business models will revolve around supporting multiple types of service with very different requirements.
That will necessitate a shift towards a very flexible, software-driven architecture, in which the physical network will be far less of a differentiator in its own right. Instead, it will be made up of commoditized servers, and of dense cell sites with advanced antennas – a combination which will enable very high data rates as well as low power IoT processes, and flexible allocation of resources via network slicing.
There is very little case, in this sliced network, for multiple single-operator RANs. A huge pool of capacity – deployed, managed and sliced by one or more MNOs, by a wholesale third party or by a cloud host like Google or Amazon – will be more economical, power efficient and flexible.
Yet it is hard to imagine this vision becoming reality in many areas, if operators and regulators cannot get past the first hurdle of sharing a conventional 4G macro RAN effectively. Or at least, while they resist the shift to wholesale and utility networks which has already been seen in electricity, gas and even wireline broadband in many countries, the new flexible platforms will be built by someone else, and the rewards of 5G will be snapped up by non-MNOs.
StarHub and M1 to share networks in Singapore:
Singapore is about to get a fourth mobile operator, Australia-based TPG, and the second and third players, StarHub and M1, are preparing for new competition by considering a network sharing deal.
The two companies have signed a memorandum of understanding to study sharing both of RANs and backhaul networks, targeting potential savings of 30% to 40% in cell site build-outs for 4G enhancement and 5G, and a consequent cost saving of 20% to 30% in capex and 25% to 30% in opex.
M1 and StarHub already share some infrastructure including antenna systems, in-building fiber and tunnel cables.
StarHub CEO Tan Tong Hai said: “We are cooperating to bring the Singapore infocomm industry to the next level, to compete not on pure infrastructure ownership, but at a higher level of customer service and innovative value creation.”
“Sharing mobile network radio elements with M1, but keeping our individual mobile core networks, will allow StarHub to provide better mobile service (in particular, mobile coverage) and still be able to differentiate ourselves.”