A 5G, community network strategy for Cuba (and other developing nations)

In a previous post, I suggested that Cuba might be able to leap over 4G to 5G wireless infrastructure using satellite and terrestrial networks for backhaul. While that would require political and policy change, it would be a good fit with Cuban culture and skills.

Before talking about Cuba, let me say a bit about wireless generations.

Each mobile technology generation used new technology and enabled new applications:

• 1G: Voice calls
• 2G: Digital data for text and sending small images
• 3G: Smartphones for low-quality video, Web browsing, and GPS
• 4G: High speed, lower latency communication for video streaming and chat and interaction with complex Web content

Fifth-generation wireless will be faster than today's 4G and latency is expected be on the order of 1 ms within the 5G network. Radios will be capable of beamforming -- rapidly switching focused beams among large numbers of devices -- and simultaneous two-way (full duplex) transmission at a given frequency. This will enable real-time applications like control of autonomous vehicles, remote medical procedures and augmented and virtual reality as well as fast file transfer and streaming and other, un-imagined applications.

Do not think of this as the evolution of the cell-phone network; think of it as a discontinuity in wireless communication to mobile and fixed users.

In addition to enabling new applications, each mobile computing generation uses different frequency bands and 5G is being designed to use very high frequencies. High-frequency radio waves enable high-speed transmission and small antennas. Being able to fit multiple small, cooperating antennas in a phone or other device (multiple inputs and outputs (MIMO) increases transmission range and speed. However, there is a high-frequency tradeoff -- low-frequency waves travel farther and are better able to penetrate obstacles like buildings and trees than high-frequency waves.

Small cell on the terrace
of a building in Bangalore

High-frequency networks will require a multi-tier architecture. With the current cellular network, phones and other devices communicate with a relatively distant base station that is connected by fiber or high-speed wireless to the Internet. Fifth-generation wireless will require many "small cell" radios that communicate with those high-capacity base stations.

Now back to Cuba (and other developing nations).

As of last year, there were 879 cellular base stations in Cuba, 358 of which had been upgraded to support 3G communication. As of 2016, 85.3% of the population was covered by 2G cellular and 47% of the population had 3G coverage. (Note that 2G coverage has barely increased since 2010 and it has been flat since 2012). If they continue rolling out 3G, it should reach the 85.3% fairly soon, but new base stations will have to be added to cover the entire population.

Upgrading from 2G to 3G requires new equipment and also more backhaul capacity between a base station and the Internet because 3G transmission speeds are greater than 2G and 3G applications use more data. For most Cubans, it would also require the purchase of a new phone. High-speed, 5G service would require much more backhaul capacity and new phones.

In densely populated areas it will be economically feasible to provide that backhaul using fiber, but fiber to support 5G capacity throughout the island would be very expensive. In many locations, satellite connectivity may turn out to be a better backhaul medium than fiber. SES Networks (O3b) will be offering connectivity using their middle-Earth orbit satellites before Cuba is ready for 5G and by the time they are ready, low-Earth orbit satellite connectivity from vendors like OneWeb and SpaceX Starlink will be available.

But what about the large number of small-cell radios that be feeding 5G base stations?

Like today's WiFi radios, they will be installed and maintained by community members and users. Cubans are known for do-it-yourslef innovation, for example in keeping old cars running and installing motors on bicycles and they have built community networks in places like Gaspar, Camagüey and Pinar del Río. Havana's SNET, is said to be the largest community network in the world that is not connected to the global Internet and there are over 8.000 amateur radio operators and over 1,400 active, self-employed programmers in Cuba.

Small cell radios will be semi-automatically configured and simpler to install and maintain than the WiFi radios used in today's street nets, but that is the easy part. Decentralized technology calls for decentralized decision making. Local people who are locally elected should decide questions like how many small cells to deploy in a neighborhood or rural area, where they should be located and how to pay for them. Would the current municipal electoral districts (from 200-3,000 inhabitants) be an appropriate locus of network control?

ETECSA would cede local control but be responsible for acquiring international bandwidth and providing backhaul over fiber or satellite from their base stations. They would also serve as consultants to local communities and could negotiate high-volume discount purchases of locally-owned equipment.

Note that I am still assuming a major role for ETECSA in spite of the fact that historically, nations like Cuba have privatized telecommunication and licensed foreign operators in exchange for investment in infrastructure. In previous posts I have suggested that vested interest and bureaucracy at ETECSA and uncertainty over control may be stifling Cuban Internet expansion. To the extent that that is the case, the new administration would have to change the organizational culture to focus on Cuba's stated economic and social policy goals -- leapfrogging current regulation and policy along with the technology. That may be wishful thinking, but if they are able to do so, they will have an advantage over nations in which private company profit trumps (no pun intended) social goals.

The technology is also ill-defined and unproven. While the standard for the first version of the 5G radio interface between a device and base station is complete, other hardware and software standards are still being developed and 5G is based on technologies that have been tested in trials, but not in large scale practice. The first deployments are not expected until next year, user and network equipment prices are not set and competing technologies like Starry may impact pricing and deployment.

Novel, unproven 5G wireless technologies (source)

In spite of this policy and technology uncertainty, ETECSA can start the ball rolling today by cooperating with and providing Internet gateways to SNET and other street nets, possibly in conjunction with support of the Internet Society and the Organization of American States. They can also learn from the experience of others like Gufinet in Spain and a variety of community networks in the US and elsewhere. At the same time, they should be following the 5G standards process directly and through their vendors, primarily Huawei and SES Networks (O3b), in order to plan for the future.

Cuba's mobile-Internet strategy?

This post is speculative, but I think Cuba may use satellite for 3G backhaul and, when the technologies are ready, leapfrog over 4G to 5G mobile connectivity and next-generation satellite. ETECSA began rolling out 3G connectivity for Cubans about a year ago and a few things have led me to believe they will continue:

• Miguel Díaz-Canel, who many expect to replace Raúl Castro, has stated that "The State will work to make [the Internet] available, accessible, affordable for all." He also cites problems and responsibilities but seems on balance to favor connectivity.
• WiFi hotspots, navigation rooms and home DSL cannot scale to bring "accessible, affordable" connectivity to all, but mobile phones can.
• During 2017, ETECSA, Cuba's government telecommunication monopoly, installed 279 3G base stations, bringing the total number of base stations to 409 and reaching 47% of the population
• Mobile connectivity is becoming available in low-population areas.
• Last December, ETECSA began routing international traffic over the O3b medium-Earth orbit satellite network and now about 5% of their international routes are carried by O3b. (O3b is a subsidiary of SES an established geostationary satellite company).
• O3b added four satellites to their constellation this month and plan to add four more next year, but they will have a much more significant upgrade when they deploy mPOWER, a new generation of satellite technology, in 2021.

The following crowd-sourced maps show Cuba's mobile rollout. (Strong signal: received signal strength indicator (RSSI) > -85dB, Weak: RSSI < -99dB).

Crowdsourced mobile coverage map, February 2017 (Source)

Crowdsourced mobile coverage map, November 2017 (Source)

Given the choice, people would prefer the flexibility, convenience and comfort of mobile or home access over access at a fixed location like a WiFi park or navigation room. Cuba cannot afford the infrastructure upgrade to make home DSL "available, accessible, and affordable for all" and if they could it would require an enormous investment in obsolete technology.

But, could they provide widespread 3G mobile? Doing so would require more base stations and more backhaul from those base stations to the Intenet. I have been told that O3b currently has a satellite-Internet gateway in Jarusco, near Havana, but my guess is that they will install others to provide 3G backhaul. This would not be unprecedented -- for example, O3b provides mobile backhaul for Digicell, which has over 40,000 LTE accounts in Papua New Guinea (PNG).

Could Cuba employ a PNG-like strategy for a portion of their mobile backhaul?

Cuba is not identical to PNG. PNG's population is only about 72% of Cuba's, but Cuba has several advantages over PNG. The area of PNG is more than four times that of Cuba and Cuba has superior, universal education, a GDP per capita about 3.5 times that of PNG and more terrestrial Internet infrastructure.

I don't know how much terrestrial backhaul capability Cuba has or what the cost of current and next-generation O3b connectivity will be, but it would be interesting to consider the possibility of free 3G mobile service. Doing so would create a generation of trained, demanding Internet users and lead to innovative, practical applications. At first, that might sound crazy, but we offer free access to sidewalks, roads, broadcast radio and more -- why not provide ubiquitous, free 3G mobile Internet?

Coming back to Earth, we might ask -- shouldn't Cuba install modern 4G technology instead of 3G?

I have long advocated a strategy of relying on stopgap measures like home DSL, WiFi hotspots, navigation rooms, street nets, El Paquete Semanal and 3G mobile service while planning to leapfrog over current technology. Third generation mobile is significantly slower than 4G/LTE, which means much less backhaul and international bandwidth is required. Furthermore, Google, industrious Cubans and other are developing applications that are tailored to work on slow connections and offline on low-cost handsets. (There were 1,432 active, self-employed programmers in Cuba as of last April).

If 3G is a stopgap while waiting for 5G wireless technology to become available, what might the future look like?

As mentioned above O3b plans to deploy their next-generation mPOWER satellite constellation in 2021. MPOWER will be a major advance. Their current satellites can link to 10 edge terminals, but mPOWER satellites will be capable of over 4,000 links each and O3b will offer several terminal models, ranging from very cheap and small (perhaps suitable for an individual cellular base station) to very large. While we may see a limited 5G rollout in advanced nations in 2019, it will not go mainstream for a year or more and will still be maturing and too expensive for either Cuba, PNG or other developing nations for some time after that, so mPOWER will be ready by the time Cuba is ready to "leap" to 5G.

Seven satellites, each with over 4,000 steerable, fully-shapeable beams

It is noteworthy that 5G terrestrial wireless is expected to be used for fixed as well as mobile access, further reducing the need for investment in terrestrial infrastructure. When we speak of 5G connectivity to fixed locations, we are moving beyond the mobile phone as a user terminal. Handheld computers work well for conversation and consuming media but not for content creation. I could have written this blog post on a laptop with a 5G connection, but not on a mobile phone.

At an mPower press conference (video), Steve Collar, SES Networks CEO asked himself a rhetorical question -- "If we wanted to deliver all of the capability that PNG would require for the next 15 years, could we do it on mPOWER without having to use any sort of meaningful terrestrial infrastructure?" and his answer to that was "yes." He went on to say that "If we can deliver the international and domestic traffic for a country on this system ... then we've got something that is genuinely unique." (Collar's comment is roughly 6 minutes before the end of the video).

Several years ago, I suggested that Cuba could use geostationary-orbit satellite Internet service as a stopgap measure until they could afford to leapfrog over today's technology to next-generation infrastructure. They did not go for that idea. Last month, I suggested that they consider low-Earth orbit satellite Internet service. This post splits the difference by suggesting medium-Earth orbit service from O3b. Since ETECSA is already an O3b customer and SES is a European company, this one may be closer to reality -- I'll save those political considerations for a future post.

=====
Update 2/27/2019

SES has published their Full Year 2018 Results. They reported a return to growth in fixed-data sales, citing a full year of revenues from O3b deployment with ETECSA in Cuba as a contributor. They also confirmed their O3b launch schedule. O3b satellites 17-20 will be launched the first quarter of this year and seven next-generation mPOWER satellites will be launched in the first half of 2021. Cuba might be planning to make significant use of these satellites for mobile backhaul and community connectivity.

Cuban satellite connectivity -- today and (maybe?) tomorrow

Last January, Doug Madory of Dyn Research reported on Cuban traffic, noting that C&W's share had increased:

Yesterday Madory reported that ETECSA had activated a new internet transit provider, medium-Earth orbit (MEO) satellite-connectivity provider O3b Networks (Other 3 billion), replacing geostationary satellite provider Intelsat:

(They have also added Telecom Italia, which, until 2011, owned 11% of ETECSA, but I will save that for another post).

O3b's MEO satellites orbit at an altitude of around 8,012 km above the equator while Intelsat's geosynchronous satellites are at around 35,786 km, therefore the time for a data packet to travel from earth to an O3b satellite and back to Earth is significantly less than to an Intelsat satellite. This move to O3b may be related to ETECSA's recent decision to offer SMS messaging service to the US (at an exorbitant price) and it will surely improve the speed of interactive applications.

That is today's situation as I understand it, but now I want to speculate on the future of Cuban satellite connectivity -- say in the early 2020s.

First a little background on O3b Networks. O3b is a wholly owned subsidiary of SES but it was founded in 2007 by Greg Wyler, who has since moved on to a new venture called OneWeb. While O3b provides service to companies like ETECSA, OneWeb plans to also provide fast global connectivity to individuals in fixed locations like homes and schools as well as the "Internet of things."

This animation was prepared by Teledesic,
formed in 1990 to provide LEO satellite
connectivity. Teledesic failed, but
technology, the market and executive skill
have changed since that time.

OneWeb plans to connect the "other 3 billion" people using a constellation of around 1,600 satellites in low-Earth orbit (LEO) at an altitude of 1,200 km and another 1,300 in MEO at 8,500 km. They are working with many vendors and partners and plan to launch their first satellites in March 2018. They will begin offering service in Alaska in 2019 and hope to cover all of Alaska by the end of 2020. By 2025 they expect to have 1 billion subscribers and their mission is to eliminate the global digital divide by 2027.

Now, back to Cuba. ETECSA is doing business with Wyler's previous company O3b. Might they also be talking with his current company, OneWeb? It takes time to launch hundreds of satellites, so service is being phased in -- might Cuba come online sometime after Alaska? By connecting Cuba, OneWeb would gain publicity, the goodwill of many nations and access to a relatively well-educated, Internet-starved market and it would enable Cuba to quickly deploy broadband technology.

As I said, this is pure speculation. OneWeb faces significant technical, business and political challenges and may fail. Politics would be particularly challenging in the case of Cuba. Both the US and Cuba would have to make policy changes, but maybe the time is right for that -- the Cuban government will change in 2018 and the US government is likely to change in 2020 when Alaska comes online.

OneWeb has established an indirect relationship with ETECSA through O3b, but other companies, including SpaceX and Boeing, are working on similar LEO projects. Might ETECSA be talking to the others?

SpaceX is particularly interesting. Theirs is the most ambitious plan and their experience as a rocket company is invaluable. Less tangibly, founder, CEO and lead designer Elon Musk is known for audacious risk-taking. OneWeb will begin with Alaska -- might SpaceX begin with Cuba? In serving Cuba, they could also serve the Eastern US and since Cuba is an island the footprint of a satellite would not be so densely populated. SpaceX would gain publicity and international political good will.

To learn more, see this survey of LEO satellite plans and related issues.

====
Update 2/27/2018

SES subsidiary O3b has gone public with their agreement to route ETECSA traffic and Doug Madory has updated the route graph (above) that he provided when he first noticed the link last December.

The following graph shows that O3b now accounts for about 5% of ETECSA's routes and, as in December, geosynchronous satellite provider Intelsat is gone. Latency over the O3b link will be much lower than it had been with Intelsat, so some users will see improved performance.

A friend tells me that the O3b ground station is about 25 miles east of Havana in Jarusco -- the location of the original Soviet Carib 1 ground station in 1974.

Satellite ground station in Jarusco (source)

Cuban international traffic shifts from satellite to the ALBA-1 undersea cable

Doug Madory, Director of Internet Analysis at Dyn Research, sent me a note on Cuba's international traffic. As you see here, on July 1, nearly all satellite traffic (blue and green) was re-routed to the ALBA-1 undersea cable:

As a result, median latency has stabilized at around 210 milliseconds:

This is good news for Cubans who have Internet access at work, school or ETECSA hotspots and navigation rooms.

There must be relatively fast terrestrial connectivity to the cable landing point at Siboney Beach. Does anyone have any information about the nature of that connectivity? Huawei is installing home DSL and WiFi -- have they also installed an inter-province backbone? Could there have been an unannounced deal with medium-earth orbit satellite provider O3b Networks?

-----
Update 7/20/2015

Huawei may be installing home DSL and WiFi hotspots in Cuba, but Doug Madory has discovererd at least one piece of Cisco equipment -- a 2800 router at the University of Havana. (I'd be curious to know how they obtained it.)

I am not familiar with the Cisco 2800, so I Googled it to get the specs. I was saddened to see that it is old equipment, near the end of its support life -- the end date for software maintenance has already passed and hardware support will end soon.