 |
GEO-LEO Hybrids: Two Mints In One
In the rough-and-tumble marketing debate over the merits of LEOs and GEOs, Teledesic asserts in an online white paper that it's not practical to separate applications that can and cannot tolerate delay (www. teledesic.com/tech/latency.html). Nevertheless, providers of hybrid LEO-GEO systems say they plan to do just that, segregating traffic bound for GEO systems from that better served by LEOs.
Motorola, for one, is taking a straightforwa
rd approach to this type of application sorting. The company plans to provide dual-use terminals--reportedly costing $750 on the low end--that let the user choose whether a given application is to be sent to a LEO or GEO satellite. The traffic will be subsequently sorted at the NOC (network operations center) using a specific algorithm for that purpose, with latency-sensitive applications sent to LEOs and broadcast and multicast applications targeted to GEOs. In the air, the traffic will move from LEO to LEO or up to a GEO and back down again.
Motorola is planning a hybrid system, says Barry Bertiger, corporate vice president and general manager of Motorola's Satellite Communications Group, since GEOs tend to limit the speed at which TCP/IP can be run. "Those applications that will be Internet types of applications, or quasi real-time or real-time applications, will always run on the LEO portion of the network, where time of flight is equivalent to fiber delay," he says. "There, we can run at 155 Mbps. GEO
s will never be able to run at those rates."
On the other hand, Bertiger notes that when a business wants "one-to-many transmission for entertainment and instructional videos, LEOs don't efficiently utilize bandwidth because they have to regenerate the signal." He envisions a scenario in which an Internet user might order a video via LEO and have that order filled via GEO.
At GEO-based Loral, which has invested $30 million in Alcatel's LEO-based SkyBridge, plans are under way to make applications-based transmission decisions at the NOC. The NOC, from the News Data division of News Corp., a direct-to-home-TV player, was scheduled to begin alpha tests in January.
Craig Partridge, a BBN principal scientist consulting to NASA, says, "It's easy to classify [unencrypted] packets according to their requirements." The problem lies in how you do it. Right now, a debate rages in the technical community over the merits of putting this type of quality of service information in a MAC (Media Access Control) laye
r header or creating a filter that will peek into the packet at every network hop.
Of course, without standards, it may be difficult to design a system for topologies that rely on multiple service providers in a given transmission path. Most of the standards work in this area is being done by the IETF's Integrated Services Working Group. If multiple providers are in a service path, the widespread implementation of a QoS (quality-of-service) standard becomes important to those satellite players sorting traffic. Another issue is that the standards work being pursued today doesn't deal with LEO/GEO differentiation, but, instead, with how packets should be handled in the face of network congestion. This effort, however, could be adapted to address LEO, GEO or even MEO assignments.
Harvard consultant Scott Bradner says there's not much of a marketing case for taking that sort of action. What users will want to know, he says, isn't whether a transmission should go via GEO or LEO, but whether it should be tra
nsmitted by landline or some other method, whichever has appropriate delay and is cheaper. He thinks satellite providers will simply use their own protocols to determine routing based on latency.
However, it's unclear how much intelligence will be built into satellite networks, and how much decision-making will be left to the user or application. As systems mature, businesses must keep this question in mind.
|
 |