HD-PLC: Data and Analysis

Continued from 'HD-PLC: Not Too Shabby'….

How'd HD-PLC do? Here's the screenshot for the office-to-hut file copy, and here's its counterpart for the hut-to-office return path. HD-PLC, at ~30 Mbps office-to-hut and ~25 Mbps hut-to-office, was better than HomePlug 1.0 Turbo, but not dramatically so. And it was comparable to, or worse than, DS2, depending on the DS2 interference filter setting. Some qualifiers on this data:
1) I didn't take it in the middle of summer, when my powerline network has proven to be most fragile. Then again, while this data is from early October, the HomePlug 1.0 Turbo and DS2 data is from roughly the same time of year, specifically late October of last year.
2) I haven't done shortwave radio or other testing yet to assess the relative interference effects of the various powerline technologies. Intellon claims that both DS2 and HD-PLC are deficient compared to the upcoming '200 Mbps' HomePlug AV in this regard. I'll need to get HomePlug AV adapters in-hand before I can validate Intellon's claim.
3) Here's more from Intellon's Andy Melder; "What we've seen (and shown at the HomePlug Conference last week), is that other non-HomePlug PLC technologies are not as robust, when tested in a real-world environment (in a typical home for example), which includes attentuation and noise sources. The best way to test this hypothesis is to stream video (multiple SD or HD streams) and observe the impact of adding noise sources, halogen lamps, phone chargers, hairdryers, etc". I now have an interference filter in the closet to block out powerline noise induced by the swamp cooler's motor and other electronics (which anyway weren't running during my HD-PLC tests). Without a large-scale multi-environment study, I can't confirm or deny Intellon's statements.

Qualifiers aside, the conundrum remains; how can vendors claim '85 Mbps', '190 Mbps' and '200 Mbps' performance for gear that doesn't come close to delivering these speeds? If you look at the fine print of Panasonic's documentation, you'll find the following qualifier: 190 Mbps "is the theoretical maximum data transmission speed of the HD-PLC standard [editor note: i.e. the PHY rate]. The actual maximum data transmission speed is about 70 Mbps. Results may vary depending on electrical conditions, network environment, and other factors. See the specifications for more information." And regarding that 70 Mbps, it's only the "maximum actual rate" for UDP transfers. The "maximum actual rate" for TCP, specifically for FTP transfers, is only 42 Mbps.

Powerline promoters will rightly point out that wireless technologies exhibit the same sort of disparity between marketed speed and reality. But wireless is already an established networking technology; powerline isn't, and I'd argue that any credibility gap would therefore hurt powerline more. And in the coming-soon era of widespread transport of high-def video, VoIP audio and other large-payload and latency-critical data around the home, any discrepancy between hype and reality, regardless of technology, will be magnified.

Speaking of high-def video, let me return to Broadcom for a moment. This isn't the first time I've heard the company 'bash' powerline. While my results bear out Palm's contention that powerline as it stands isn't yet ready to reliably transport multiple high-def streams around consumer's homes, and that the diversity of incompatible protocols (DS2, Panasonic, and HomePlug) will cultivate consumer confusion, I will suggest several 'but's…..
1) Feedback from broad deployment of powerline, coupled with additional silicon suppliers entering the market, will likely consolidate the current protocol profusion and improve the techology's speed and robustness over time.
2) Mine is something of a worst-case configuration; I didn't measure transfers between my office and closet, for example, which likely would have been faster, and newer homes (or those with less complex power topologies) might also experience a performance and reliability boost.
3) Is the need to simultaneously stream multiple high-def streams really going to be the predominant usage environment in the home, especially over the next few years?
4) If so, the bitstream reductions delivered by leading-edge video codecs like H.264 (aka MPEG-4 part 10, aka MPEG-4 AVC) and VC-1 (aka Windows Media Video 9 and followons) will make the implementation of this environment more feasible than continued reliance on archaic legacy codecs like MPEG-2.
5) Finally, I think Broadcom's bashing has at least 'a bit' to do with the fact that powerline isn't currently a part of its product portfolio. Don't you?

One interesting tidbit in Broadcom's talk, which I'd welcome your feedback on, is the presenter's contention that powerline's single frequency band reliance would result in performance degradation when, for example, multiple apartments within a complex or multiple homes within a neighborhood simultaneously went 'online'. My experience suggests that powerline data has limited effective 'reach', and I also assumed that multi-dwelling interaction potential would get squelched by circuit breaker boxes, meters, transformers, etc. High voltage AC experts out there, does Broadcom's contention have merit?

At the Intel Developer Forum last month, powerline networking session presenters confidently stated that gear based on HomePlug AV technology would be in consumers' hands in time for this year's Christmas shopping season. I certainly hope so….the hype-to-reality time lag has already been unacceptably long. I hope to have adapters in-hand within the month and will report back my findings as I get them. Until then, as always I welcome your thoughts.