I have been experimenting with powerline carrier modems for many years, going back to the first 10 Mbps products of more than a decade ago. These do exactly what you think - turn the electrical wiring in your house into a virtual LAN cable. In one sense, we might even call this wireless, since we're using somebody else's wire.
But, as you might also guess, this is an endeavor fraught with complications. AC wiring is most certainly not designed to carry data; based on some crazy wiring I've seen in a number of homes over the years, it often can barely carry electricity! While AC wiring itself is standardized, two phases are normally present, and reversing two of the three wires involved (hot, neutral, and ground) is not uncommon. Even low-speed (e.g., X10) applications can maddeningly live somewhere between less than reliable and time to give up.
So, when a new crop of 500 Mbps HomePlug-compliant products started to hit the market, it was time to think about testing a few of these to see what they could really do. I was expecting disappointment, based on what I noted above - the technology is fabulous, but the PHY of house wiring can be even worse than the PHY of the world of 802.11.
I ended up with two products to test: ZyXEL's PLA4205, and Netgear's XAV5501. These are identical in concept - a small box that plugs into AC (the Netgear is even three-prong) and has an Ethernet port on the other end. Setup couldn't be easier for both: plug both ends in and you're done. Indicator lights show Ethernet and powerline connectivity, and that's that.
Except for performance. For this test I used some very high-end gear - an Ixia VeriWave WaveTest 90 chassis with just an Ethernet card as one end of the link, and WaveInsite as the test suite. WaveAgent software was installed on a PC on the other end of the link. And, in the middle, the powerline modems under test. The test load consisted of upstream and downstream HTTP traffic of sufficient volume to saturate the link.
The first step was to try the test with both modems connected to the same power strip. This would likely yield the best possible performance. Three runs of each product set produced the following results (goodput, in Mbps):
ZyXEL 235.352 242.172 231.645 Average 236.390
Netgear 235.132 189.094 120.252 Average 181.493
The variability in Netgear's results is hard to explain, but certainly could have been the result of glitches at the physical layer due to, well, AC wiring. It would be wrong to reach any epic conclusions from this test; I was really just looking for ballpark capabilities. And the performance here really wasn't all that bad - certainly well within in 802.11n territory.
The bad news, though, is what I found when I moved the far end of the link to a distant AC plug in the house:
ZyXEL 7.326 9.540 9.801 Average 8.889
Netgear 15.888 16.652 17.297 Average 16.612
Which is kind of what I expected. The problem with powerline carrier isn't that it can't work, but rather that it is likely to result in poor performance without the user really being able to understand why (kind of sounds like wireless, doesn't it?). A diagnostic tool of some form analogous to the now-common and sometimes-very-inexpensive spectrum analyzer would be useful here, but I suspect that test equipment along these lines won't be seen at popular prices for some time if ever (due to limited demand), although it might be worth looking into a software tool based on these modems themselves. I am a big fan of building advanced diagnostic and performance analysis capabilities directly into the firmware of devices like these, and I hope such becomes more common going forward.
The only other points of note is that there was always a 3 or 4:1 ratio of downlink from the test chassis to the far end) to uplink goodput, and zero packet loss in every test case.
Given the relatively poor results of the distance test, it would be hard to recommend powerline products in demanding applications. They are not a good substitute for 802.11n, and regardless are by definition not mobile. I would consider them for same-room links, perhaps in a home theater or home-office setting, but I suspect these products, despite a very long gestation period for the underlying technology, will not be going mainstream anytime soon.