Brian's Brain

This blog post references my cover story 'Home Transportation: Benchmarking Powerline, 802.11, and Ethernet' in EDN's August 2, 2007 edition. It's one of a series of web addendums to the print writeup.

My hands-on project encompassed three physical transmission mediums for network communications; Cat 5e cable (in 100 Mbps and 1 GbE versions), my home's power grid (in HomePlug 1.0, HomePlug 1.0 Turbo, HomePlug AV, HD-PLC and UPA flavours), and the air (in 802.11a, 802.11b, 802.11g, and both 2.4 GHz and 5.8 GHz draft 802.11n variants). As Maury mentioned in a recent editorial, at least two other networking physical layer options exist; coax cable and phone wiring. There are several reasons why I didn't include these additional alternatives, beginning with the pragmatic factors of time and space.

As you've probably already ascertained from reading the main article, planning, setting up and running the project, along with analyzing the results, took a ton of work. Adding even more technologies to the mix frankly might have resulted with me in a straightjacket. And as it was, as you'll see in comparing the print and web versions of the main article, we needed to do a last-minute cut of some of the material originally intended for print, putting it instead in a web-only sidebar....not to mention the planned-from-the-beginning web-based addendums such as this one!

Neither coax nor phone wiring would have been compatible with this particular home's test bed. I only have phone plugs in five rooms (including the garage!); notably, there's no telephone jack in the living room. Similarly, coax only runs from an over-the-air antenna on our roof to three rooms in the house, and it seems to have some issues at various frequency points. My coax and telephone wire topology limitations are typical of those in many homes, and retrofitting a residence with additional wiring spans is at least as painful as alternatively running Cat 5e or more costly Cat 6 cable from the router to the desired endpoint(s).

HomePNA v1.0 is only specified to run at up-to-1 Mbps rates, and v2 only runs at 32 Mbps peak rates "with throughputs approaching 20 Mbps" (which is why HomePNA is moving to coax as its next-generation physical medium). And coax cable, while bandwidth-rich, is pretty much a North America-only phenomenon; when crafting an article, I always need to keep in mind the substantial percentage of EDN readers that reside outside the U.S. Finally, consider that even if coax or phoneline runs to your desired room, it may not enter the room at the exact desired location. What will you do, for example, if the jack is on the opposite wall from the one where the equipment to be networked resides? Considering the limitations of coax- and phoneline-based networking, particularly as they relate to retrofits of existing homes, I've tended to lean towards Cat 5e cable as the preferable 'backbone' technology approach....that is, unless you can alternatively leverage the AC jack that your gear's already being powered from, thereby explaining why I also included powerline networking in this study.

One final comment in closing; while my project tested generic 802.11, I consciously chose to 'pass' on other wireless approaches such as UWB and proprietary 802.11 variants such as Amimon. This is because I wanted to focus on technologies that could transport generic TCP and UDP traffic, therefore being capable of use in a diversity of networking applications (versus only being applicable to specific protocols such as wireless USB and audio/video 'beaming').