Sunday, February 1, 2009

Coding in space

Local telecom industry veterans, particularly those who got a good dose of information theory in their EE degrees, understand why Andy Viterbi won his Shannon Award. All would know that Viterbi developed what (proved to be) the maximum likelihood decoding scheme for convolutional codes, which remained the state of the art until turbo coding came along.

Most would know that every GSM cellphone (let alone CDMA cellphones) includes a Viterbi decoder. Some might know that one of Qualcomm’s first products (and I believe its first semiconductor product) was a Viterbi decoder chip.

A few would recall that Klein Gilhousen led the group that designed the LMP (Linkabit Microprocessor) which was used to implement the Viterbi algorithm for the Dual Modem.

What might not be known — I didn’t know until I researched the back story for the book — is that before cellphones, chips and the dual modem, the first real application for the Viterbi decoder was for NASA deep space probes.

In the late 1960s, NASA was planning two competing families of space probes for mankind’s first visit to the outer planets. The communication challenges were among the most daunting in history, where (due to the inverse square law) the signal strength a billion miles away was faint, and power (from nuclear decay batteries rather than solar power) is minimal.

Theoretically, the application of Claude Shannon’s ideas (and coding theory) could provide as much as a 10 dB signal gain. MIT-trained academics worked with two different divisions of NASA, both using coding theory:
  • Pioneer 10 and 11: Codex of Newton, Mass. worked with NASA Ames to apply Fano decoding
  • Voyager 1 and 2: Andrew Viterbi, Joe Odenwalder and others — first as a UCLA contract and then as a Linkabit contract — worked with JPL to use concatenated decoding and the Viterbi algorithm.
My research on this led to chapters on MIT and NASA for the book, as well as changes to other chapters. However, I’ve stopped quoting an ETA for the book given (as a part-time job) it’s coming very slowly.
In the meantime, two years ago I wrote a paper for the Journal of Management Studies, in a special issue on technology commercialization. The paper, entitled “Commercializing Open Science: Deep Space Communications as the Lead Market for Shannon Theory, 1960–73,” was published in December after several rounds of review.

Errata (Feb. 4, 2009). I received an email from Andrew Viterbi with a correction to the paper that will be applied to the book:
the first Linkabit customer was the Naval Electronics Laboratory Center (NELC) in San Diego… NASA was the second, but never became a large customer.
Most universities subscribe to the journal, so the paper is available as online from a university IP address (DOI: 10.1111/j.1467-6486.2008.00807.x). Normally I post a PDF of the pre-publication Word document, but there were major changes after submission during editing the page proofs. However, I would be glad to email the published PDF to anyone who asks (at ).

Graphic credit: Brewster Rocket comic from Sunday, February 1, 2009. Note to Tim Rickard: Pioneer 10 was launched first, but Voyager 1 was actually the first manmade object to leave the solar system and the most distant space probe. So your aliens would first see the Golden Record rather than the Carl Sagan memorial plaque.

No comments: