Flexibility defines the coming generation of broadband satellites
The next several years will see a rapid progression in deployment of broadband satellite services. Transitional satellite payload designs will be based on conventional transponders. But eventually, on-board switching and processing payload designs will exploit broadband satellite and frequency resources. The key to success? Deliver a cost-competitive service and do so in a manner that is flexible enough in design to keep up with a very dynamic decade of telecommunications changes.
Satellites that will successfully complement terrestrial telecom networks must deliver services that are better provided from space.
Traffic modeling performed by TRW (the authors’ company) has shown that successful designs will be those that provide the largest amount of traffic capacity while also providing flexibility to allow use of the bandwidth in a manner that is not fixed but market responsive. This article will show how the progression of satellite capabilities has led to an entirely new emerging generation of satellites that process signals in space to achieve both capacity and flexibility.
Response to market shifts
Future satellite systems will have market demand similar to terrestrial telecom — more bandwidth, improved functionality, lower price. The broadband systems with the greatest capacity and simplest terminals to use will have a market advantage. But there are important differences. This is where the flexibility of broadband satellite services becomes important.
Consumers will become accustomed to always-on, high-speed data connections as part of daily existence. Unfortunately, terrestrial broadband links, whether wired or wireless, typically have shorter range than the voice-based systems they will replace. This can lead to pockets of broadband isolation even in relatively urban areas, let alone areas where lowdensity population or economic deprivation render dedicated terrestrial systems uneconomical. For these areas, satellites appear the natural broadband choice.
Satellites have always been providers of true universal service — signals from space do not care whether they land on urban or rural antennas. What differentiates the coming broadband services from past satellite systems is that the target price will be comparable to terrestrial delivery, not at a premium. This means that satellite-delivered broadband has the potential to reach beyond hard-to-reach areas to become a true mass market service, just as direct-to-home satellite TV has become.
The large geographic scope, however, means that each geostationary satellite will serve multiple countries with a wide range of bandwidth demands and service expectations over its lifetime. For a time, some countries may have a high demand for Internet access, while others may be heavy generators of video multicast or corporate extranets. Not only will these change gradually with the economic climate but there will also be more rapid changes on an hourly basis, as the day sweeps across a continent or special events suddenly shift demand. Only by employing the latest bandwidth-on-demand techniques can satellite designers ensure that such systems will be cost effective over their entire life cycle.
What to expect
In the first, a transitional technology mix of DTH-like data delivery is already in limited consumer use in products such as DirecPC. Users share a limited bandwidth satellite downlink that may provide faster Web downlink surfing, provided that there are not too many people online at a time. User uplink transmissions utilize the conventional telephone modem path, which restricts the services that can be supported.
These first generation satellite broadband systems are already giving way to two-way services that don’t rely on a modem connection. Besides simplifying the connection, two-way services such as Gilat-to-Home provide more dedicated satellite bandwidth and the ability to send, as well as receive data at faster rates. This allows personal Web hosting and the basis for limited interactive service.
Both of these early generation broadband satellite systems rely on existing Kuband (11 GHz) satellites, which allow them to be deployed quickly but are not the most cost-effective or flexible resources for long-term application. For the near-term, however, these can be profitable uses for excess transponder capacity and are likely to have appeal to geographically isolated and early adopter market segments.
True broadband performance shows up in the third generation systems, which are already in production and will first come to market in late 2001. Unlike the transitional first and second-generation systems that rely on traditional Ku-band transponders, these satellites use Ka-band (18-30 GHz) for satellite to-and-from ground transmissions. The higher frequency of the Ka-band systems allows easy implementation of satellite spot beams, which when combined with channelized switching provides a capacity of a single Ka-band satellite that can easily exceed that of a Ku-band satellite by a factor of four or more. With little increase in system cost, the number of potential customers reaches mass-market proportions and the effective cost per circuit drops dramatically. With such systems, Ka-band allows satellites for the first time to provide access-oriented circuits to an entire continent at prices comparable to terrestrial delivery in cities. These will find broad residential application where access services are most price sensitive.
The use of spot beams comes at a price. In order to provide the flexibility required by the long life and broad geographic scope, a satellite must be able to shift capacity among beams. This requires onboard switching, which gets more complex as the number of beams increases. Still, a capable on-board switch can often improve the effective capacity of a satellite by an order of magnitude for many traffic models, making it well worth the early expense during development.
A fourth generation emerges
The first three generations of broadband satellites are often termed access technologies — traditional transponded, or bent pipe, satellites, that are focused on delivering a broadband connection between users and the Internet much as modems provide connection through terrestrial wires. In both cases, each circuit handles one user at a time and the routing of data is performed in a central facility. If the traffic from many users is combined in a satellite terminal to fill the data channel continuously this can be a very efficient system. Unfortunately, as satellite terminals get smaller and cheaper, they become more single-user dedicated and the circuit takes on a bursty traffic pattern. The power-limited satellite down-links are not completely filled with data and therefore the capacity is used less efficiently.
As these third generation Ka-band transponded systems approach market rollout, the fourth generation approach is being readied to address this downlink efficiency problem. By moving the traffic routing function from a central ground facility up to the spacecraft, system designers can combine the traffic from many bursty users before it gets to the satellite downlink transmitters — dramatically improving efficiency and with it the ‘billable Bits’ that result in higher revenues for the system.
This efficiency can provide higher band-width-on-demand availability and quality of service (QoS) to more users because it dynamically and automatically allocates unused capacity as needed. Customers pay only for what service they need at the moment they need it, at whatever quality level they choose.
The processed payload design is almost identical to the transponded design. However, the on-board circuit switch found on transponded designs is replaced by onboard demodulators, a switch and a set of downlink modulators. In addition to highly efficient use of the downlink capacity, the use of on-board demodulation leads to improved link performance, which reduces both user terminal size and transmit power requirements, since the signal is regenerated before downlink noise is added. In addition to isolating the uplink and downlink interference, onboard demodulation also provides a high degree of access security, ensuring that rogue terminals do not rob power from the downlink.
The heart of the processed broadband satellite is a fast packet switch. The irregular incoming data streams are pooled and distributed to the output channels, ensuring that the downlink remains maximally loaded under all traffic conditions. This “statistical multiplexing” — the multiplexing of several partially utilized up-link channels (bandwidth allocations) into downlink channels which fully utilize the capacity for user data — is an extremely important capability since most services present statistically bursty data traffic to the satellite uplinks.
For network operators, transitioning from a traditional transponded to the emerging generation of processing architectures raises the question as to which architecture maximizes throughput and revenues. Table 2 summarizes the results of a TRW analysis that quantified the throughput potential of transponded, circuit switch, and fast packet switch architectures to determine the raw capacity achievable and how efficiently the capacity is used (average throughput) for multimedia traffic in mesh networks. In the study, TRW defined a traffic model made up of various services, including broadcast-quality video, videoconferencing, voice and Web surfing. The results show the improvement that on-board processing provides to the average throughput, a measure of deliverable user data, under these mixed traffic conditions. The raw capacity and average throughput (billable bits) of the processing architectures are significantly larger than the transponded architecture as a result of the frequency reuse possible without a terrestrial infrastructure. The transponded architecture requires extensive ground interconnection to deliver full-mesh connectivity. Even then, with only one beam per satellite, frequency reuse is not achievable. This limits the throughput and revenue-generating capacity of the transponded architecture. The circuit switch architecture enables frequency reuse by employing multiple beams. Inter-beam connectivity is only achieved via ground gateway interconnections. The advantage of the circuit switch over the transponded architecture comes from the frequency reuse gains, and its ability to route bandwidth to other smaller geographical regions. These architectures are ideally suited for traditional broadcast television services, but do not efficiently transport multimedia applications from a large number of geographically distributed users. The most flexible and efficient approach is the fast packet switch architecture, which allows a user in one beam to connect directly to any other terminal in the network without investing in significant ground infrastructure. The fast packet switch also takes advantage of the bursty nature of variable rate multimedia sources to most efficiently utilize the raw downlink capacity to achieve billable user throughput. The packet switch architecture delivers a substantial advantage over the other architectures in terms of available billable throughput and ground infrastructure requirements, enabling the network service provider to market and sell gigabits of capacity and very profitably support a variety of traffic scenarios.
In addition to the capacity advantages that processed payloads provide, on-board switching allows the capability to dynamically configure the satellite connectivity to serve any number of specialized uses simultaneously. This allows multiple services to be offered with an ability to shift the satellite bandwidth allocation as market demands change. A nimble fast packet switch can handle basic Internet protocol (IP) frame relay, ATM services, point-to-point and multicast video distribution, local-into-local retransmission, local content insertion into national feeds, corporate mesh virtual private networks (VPN) and a host of other specialized broadband services.
The primary difference between the processed and transponded payloads is the switch. The recurring cost associated with the complex digital implementation is a bit higher than for a simple circuit switch, but is a relatively minor cost increase when compared to the investment required for the remainder of the satellite and the ground network infrastructure. The additional capacity or billable bits far outweigh the small incremental cost increase.
Broadband processed satellites are here
With compelling demand and recognized market advantages, several fourth generation on-board processing designs are already in development and are expected to come to market by 2003. With the market set to move through several generations of broadband solutions in just a few years, customers will benefit from rapid improvements in performance, reductions in price and a wave of new interactive service capabilities that will make broadband service practical everywhere. Having learned its lesson from single point offerings, this time the satellite industry is prepared for success.