Intelligent arguments

Over the last few years, satellite system designers and operators have engaged in an increasingly heated dialogue. The topic? On-board processing and switching for broadband satellites.

The debaters break neatly into two camps. In one corner: the “yes, on-board switching is the future” camp; and in the other corner: the “no, it’s a bunch of over-programmed hooey” camp.

Naturally, the two sides have much to say.

The concept of on-board switching is not entirely novel. For 10 years, the U.S. government has employed on-board switching for narrowband applications on its ACTS and Milstar satellite programs. Mind you, these are narrowband programs, not broadband.

On paper and in limited practice, onboard switching appears a mighty challenger to ground-based switching and processing — particularly for broadband satellite networks. Most Ka-band systems will feature first generation on-board switching.

But drop the phrase “on-board switching and processing” in casual conversation among operators and take cover. Passions run high.

If you elect to put the smarts on a satellite and launch it and something goes wrong, some critical component takes a dirt nap, what then?

In a traditional bent pipe, transponded satellite network, all the smarts are ground-based.

The complexity of on-board switching is daunting because those components responsible for routing and processing digital data chunks will be far, far above the Earth in a LEO (500-1,000 kilometers) or GEO (36,000 kilometers) orbit, depending on the network architecture. if something fails catastrophically with the on-board components, it will be a long journey, even for the stalwart engineer. To avoid funky Stanley Kubrick-type scenarios that include orbiting engineers with tool boxes, most proposed Ka-band satellite systems have plans for full redundancy, just in case.

On-board processing makes the best of full-mesh connectivity, single-satellite hops, bandwidth on demand, frequency reuse.

But again, bent pipe technology can achieve almost all of the same things.

The goal of on-board processing: exploit resources. Make the best of spot beams and beam switching and do it with a splash of bravado.

The problem, of course, is how fancy do satellites need to be? And if on-board switching is so deliciously splendid, as the proponents claim, why isn’t it accepted as gospel?

It’s not just the distance between an engineer and a payload that’s worrisome. It’s about money. Money. Money. Money.

Why pay Ferrari prices when a Volkswagen will do?

Adding on-board processing is expensive, and the more complex and sophisticated the design, the higher the price. Proponents of on-board switching claim that this higher price is recouped through the higher throughput. The cost per bit of information goes down, and operators can haul more “billable” bits with smarter payloads.

The bent pipers are far less bullish, naturally, and proselytize that the risks of onboard processing far outweigh the advantages. “If it’s not broken, don’t fix it,” is the battle cry.

In the grand scheme, there’s probably room for both designs. So, bent pipers and on-board switchers, put the gloves down. You’re both right.

Just remember the maxim: Don’t let technology exceed application.

On-board processing may be risky. But it may work.

Bottom line: Quit arguing theory. Build the fancy satellites, if operators are willing to ante up, let them. We could use the bandwidth anyway.

Leave a Reply