Over the last two weeks, we have seen a lot of news coverage and resulting public discussion and debate on the use of satellite technology to provide ubiquitous broadband, especially in areas where fibre is difficult to deploy, as it is either financially unviable, or too difficult to trench. Services including those provided by Avanti and its Hylas-1 Satellite, or the Google-backed O3b Network, will suffer too many drawbacks to provide adequate, future-proof super fast broadband, even as a solution of 'last resort'.
This is primarily due to an inherent lack of capacity, lack of bandwidth, poor latency, sub-optimal availability, and poor economic value both to end operators and subscribers. Let’s explore each of these individually:
The fact that the satellite covers an entire territory leads to a limited total capacity as the spectrum of one single spot is unable to be re-used. Even though next-generation satellite services support 8 spot beams, once these are deployed and fully consumed, you will need to launch another satellite to meet increasing demand. Furthermore, the increased number of satellites in low orbit increases congestion, leading to increased chance of collision, increased coordination and increased risk of satellite-to-satellite interference. The capacity required to provide high speed broadband to 1% of 30m households represents at least 150Gbps, and it is doubtful that geo-stationary satellites will be able to offer this kind of capacity for quite some time.
Satellite downlink speeds -- even in the Ka band – are limited to 10Mbps. In fact, it is doubtful that most users will get a constant 10Mbps throughput with these recent announcements, given that residential services are typically contended at a rate of 20:1 - 50:1. Even if capped at 10Mbps, with a contention of 50:1, end users will only get about 256kbps download along with a very slow upload.
Analysts including Ian Fogg from Forrester have already picked up on this. Geostationary satellites are typically 35,000km above the meridian. A signal travelling at the speed of light would take about 250ms from the user to get to the satellite and to the ISP or another user (without factoring in signalling or routing/switching delays). In an IP world, this theoretical minimum doubles to 500ms before an answer is received. In the “real” world with real delays of network resources, the normal latency with satellite services is expected to be in the range of 500-700ms, with a round trip time (RTT) of up to 1400ms.
Newer generation satellite networks such O3b state that they get around this by orbiting at 8,000km. This would theoretically bring down latency to 350ms RTT, on average. This is still over and above what most modern broadband services require. This places a tremendous impact on the customer experience and today’s next-gen services consumed by subscribers, for example:
VoIP (Voice over IP) or Videoconferencing:
Latency is very irritating and debilitating with interactive applications, such as VoIP, or videoconferencing, whilst free market applications such as Skype tend to behave unpredictably and fail. VoIP callers usually notice roundtrip voice delays of 250ms or more.
ITU-T G.114 recommends a maximum of a 150ms one-way latency. Since this includes the entire voice path, the end user’s own network should have very low transit latencies. Most network SLAs specify a maximum latency 45 – 65ms. Clearly, VoIP over satellite does not perform well.
Mutiplayer online gaming:
Photo Courtesy of Dekuwa
Most first person shooters require <100ms pings, and some more complex 3D online multiplayer games require closer to 60ms. Users of Avanti's satellite service will find it difficult to gain access to Xbox live or Playstation online gaming rooms. New motion detection games based on Xbox Kinect or Playstation Move will be virtually unusable.
Quality of service and the customer experience is critical when trying to increase the adoption of super fast broadband. What consumers are really asking for is a green light to run HD videos on applications like BBC iPlayer or Hulu. They want to experience stutter- and jitter free Skype audio and video calls. Customers want to play online, and they want to interact. Stating these new satellite-based networks will be able to provide 10Mbps and super-fast broadband is misleading to consumers, as they will not be able to run any of the above services effectively (assuming they actually get 10Mbps in the first place).
The availability of satellite is only 99.5% in the Ku band and a mere 99% in the Ka band. Increasing modulation to obtain more capacity comes with a price tag of reduced availability. This leads to frequent reductions in available bandwidth, or even broadband disconnections during heavy rain or snow. All you have to do is look at how satellite TV performs under these conditions (pixelisation, image degradation, etc.). Furthermore, in the event of an outage, the mean time to repair is likely to take days, if not weeks, leaving both consumers and enterprises without any broadband service for an extended period of time.
The initial investment for new satellite infrastructure and receiver technology is significant. Given the low total capacity, it will be difficult to achieve economies of scale for reduced pricing on customer premises equipment. Increasing capacity will require a continuous battery of costly satellite launches and increased congestion and interference in the lower orbits where they operate.
It is my opinion that satellite broadband is not efficient for next generation superfast broadband. Compared to *any* solution, the OPEX and CAPEX costs for the poor level of services offered associated with a typically and predictable low take-up rate will not offer a viable business case for carriers, ISPs, and enough value for consumers.
Stated even more plainly, satellite broadband’s sky-high aspirations for bridging the Digital Divide, booming consumer & enterprise demand, and boosting operators’ business models are fated to plummet back to earth.