AdvancedTCA lays groundwork for Iridium NEXT
Interview with Hermon Pon, Iridium Communications, and Michael Lohman, Hughes Network Systems
Iridium NEXT, a new iteration of the Iridium satellite constellation, is currently in development and will provide higher bandwidth per user, hosted payload capability, and other improvements. At the heart of the Iridium NEXT ground architecture is the Access Network Controller (ANC), powered by 40 Gbps AdvancedTCA (ATCA). In this xTCA & CompactPCI Systems interview, Hermon Pon of Iridium Communications, and Michael Lohman of Hughes Network Systems discuss upgrades to the Iridium ground architecture, and how ATCA is helping roll out the NEXT constellation. Edited excerpts follow.
PON: We currently have an operational network of 66 satellites in Low Earth Orbit (LEO) that operate in six polar planes, and then we have some orbital spares. The satellite network has cross-links, something unique so the satellites can communicate with each other. These satellites were all launched in the 2000 timeframe, and even though they are providing high-quality service today, they are nearing end of life.
The Iridium NEXT project is about designing a new generation of satellites that will be a one-for-one replacement of the current constellation. In addition to replacing these satellites, a bunch of things are being upgraded. There are certain sites we call feeder links that terminate into the ground network, and those are all being upgraded as well. Also, we are upgrading the ground network to support Iridium NEXT, which is really the nature of the Access Network Controller (ANC).
Q: Generally, where does the ANC fit into Iridium's gateway ground architecture?
PON: We have feeder links that are located in a number of areas of the world that bring traffic to the ground, and then we have an MPLS network on the ground via facilities we get from professional carriers. So, from the teleport site traffic is brought to one of our gateway sites, which are pretty large and quite capital-intensive. We have all of our equipment there for terminating to the terrestrial network, billing systems, telemetry, monitoring; all that goes on at the gateway sites.
There is one logical ANC at each gateway site.* Our ground network was originally based upon a GSM network, so we have a commercial switch that is provided by Ericsson and the ANC is equivalent to the Base Station Controller (BSC) or the Radio Network Controller (RNC) function in a GSM or 3G network (Figure 1). It interfaces to things like routers, Ethernet switches, and the Ericsson switch. It also interfaces to our network management system, which is based upon server farms that are typically running things like Oracle databases.
Our single ANC handles the standard handset service for all of the voice users we have in the globe, and we have provisioned 700,000 active users in the Home Location Register (HLR). In terms of things we need in the ANC, we need a lot of high-capacity control plane processing. We need lots of media processing because we have voice codecs, so we have to convert from the Iridium codec that is used over the satellite link to traditional landline codecs. In addition, we have data services that have error correction. All of that is implemented in the media blades that are on the ANC.
Q: What are some of the capabilities we can expect from the NEXT system, and what led to the selection of AdvancedTCA as the compute architecture for the ANC?
PON: The most important capability is that the Iridium NEXT satellites will actually continue to support all our existing user terminals, handsets, and so on; they will be totally backward compatible. In addition, we will be introducing some higher bandwidth air interfaces, which will allow us to increase the available bandwidth per user. The highest speed we offer today is 120 Kb; with NEXT we go up to 400 Kb. Even with our voice terminals we expect to be offering voice and data at much higher rates. We have to have enough capacity on the ground to continue to add new users, but then the bandwidth per user is also going to increase by a factor of two or three. That is why we need a very high-capacity platform for the ANC, and why we went down the ATCA path.
When we started evaluating various computing platforms, we were looking for fault tolerance and the ability to plug in hardware or do in-service software upgrades. There is only one ANC, so it has to be operational all the time - we cannot take it out of service. We also need scalability so that we can increase capacity by either adding new server blades or adding additional chassis. That led us to choose ATCA for the ANC, and as part of that project we contracted with Hughes Network Systems. They were actually working on ATCA at the time, and did the actual software development and integration before delivering it to us.
LOHMAN: The ATCA platform was chosen by Hughes for the ANC due to its High Availability (HA) feature set. Hughes also had years of experience working with ATCA, so it was natural for us to use it. Furthermore, the ATCA eco-system has a rich set of card types and modules. We use very high-end Intel processing cards, high-speed fabric modules, very dense DSP modules, as well as legacy STM-1 TDM modules for the ANC application. Everything we needed was off the shelf.
PON: The ANC today is a dual-chassis rack of equipment, and it actually replaces something that was five racks of equipment (Figure 2). We went from five racks of equipment to one with ATCA. Another reason we chose ATCA is the ability to add additional chassis as needed in the future without having to replace everything else.
The other thing we are doing with ATCA is standardizing it as the computing platform of choice for all of our gateway nodes. The ANC is just one of our gateway nodes, and we have other nodes in the gateway that support a service we call Push-to-Talk (PTT). That is built on an ATCA platform too. So we like ATCA because of the commonality of maintenance practice, our operators all know how to use it, we can leverage it and reduce the number of spares we need, and then in addition we have the ability to scale and add more as we need it.
Q: During the one-for-one satellite replacement process, there are going to be a lot of incremental upgrades. How does ATCA help facilitate the Iridium NEXT rollout?
PON: That is one of the reasons why the ability to do in-service software upgrades is pretty important. We do have very rigid control on what we call the interface control documents, or the interfaces from the ground towards the satellites; those are not allowed to change. So we could change something on the ground, but it should be transparent to the satellite; we do that regularly, and have been doing that for the life of Iridium.
With the previous platform that was in the ANC, we had to do a hot cutover to ATCA, so we went from all the traffic on the old system switching over to the new system on the order of a few seconds. We did not have to make any changes to the satellite. The big complication in development was making certain that we did not change any of the interfaces that go toward the satellites in an unplanned manner. With NEXT, the plan is to make the ground systems kind of dual mode to support the ability to know when we are talking to a NEXT satellite or an older satellite. That is in development right now.
Q: What are Iridium's requirements for compute platforms of the future?
PON: Obviously the continued evolution in speed. Anything that improves performance and lowers the price is important.
A really important thing for us is longevity in the sense that our biggest investment is really in the satellites, and it is very important that we maintain this compatibility on the ground. We cannot afford at some point to say, "ATCA, we are not going to have that anymore. It is going to be 'ATCA-some new version' with a new chassis and you cannot buy any more blades on that." We are betting heavily on the fact that if ATCA changes, it will be evolutionary rather than some revolution where all of the vendors decide not to make it anymore. That is why ATCA was important to us: the fact that it looked like it was going to be around for a long time and that it was very evolutionary. There was a lot of multi-vendor support so we were not tied to SUN or HP blades, and it was designed to be open so that we could mix and match different vendors and have the ability to keep evolving the ground system slowly, rather than being forced into an upgrade because we chose a vendor that decided to come up with something totally new.
Q: How viable is ATCA as a compute architecture for other future satellite systems?
LOHMAN: Hughes has deployed solutions based on ATCA for several mobile satellite operators worldwide, and we plan to continue to do so for the foreseeable future. In fact, in some of our most recent work, we have been migrating operators from other platforms to ATCA.
Iridium Communications | iridium.com
Hughes Network Systems | hughes.com