Bond. Channel Bond

11:25 AM -- As we mentioned yesterday, Get of Norway has apparently raised the Docsis 3.0 speed bar via a trial that, it claims, pumped out downstream bursts of 1.4 Gbit/s. (See Get Tests 1.4-Gig Speeds.)

A Get spokesman has since responded to some questions, and confirmed that the lab test involved the bonding of 32 8MHz-wide downstream channels -- definitely outpacing the 1.17Gbit/s Kabel Deutschland GmbH 's produced in a test that bonded 24 8MHz channels. Cisco Systems Inc. (Nasdaq: CSCO) modems and cable modem termination systems (CMTSs) were involved in both trials. (See KDG, Cisco Hit 1Gig Cable Modem Speeds .)

A source familiar with EuroDocsis notes that the numbers suggest Get's trial achieved "real" throughput of about 43 Mbit/s per channel, compared to the theoretical maximum of 55 Mbit/s per channel that EuroDocsis is capable of producing.

Nothing's been decided yet, but the next integrated D3 chipset is expected to bond 16 or 32 downstreams, so this partly explains why we've seen a battery of tests recently that push the Docsis 3.0 envelope. (See The Ultimate Cable Modem and Korean ISP Bonds With 800-Meg D3 Test)

But what's the application? IP video is the obvious candidate, but even 32 bonded downstreams still seem rather extreme.

Last October at the Society of Cable Telecommunications Engineers (SCTE) Cable-Tec Expo in New Orleans, cable engineers suggested that the number of channels required for a IP video simulcast could fluctuate. It's possible that just four will be needed during the early phases of that migration, followed by eight as standard-def and hi-def channels are added. But it's also been estimated that a full offering, with 50 percent of the homes passed getting that simulcast, might require 16 channels. (See Cable-Tec Expo: What's the Magic IPTV Number? .)

So a 32-channel bond will easily cover that, and provide plenty of headroom.

— Jeff Baumgartner, Site Editor, Light Reading Cable

Be the first to post a comment regarding this story.
Sign In