Powerline Networking: Interview Insights and Additional (Potential) Issues

If you haven’t already done so, I’d encourage you to check out my recently published interview with DS2’s vice president of technology and strategic partnerships, Chano Gómez. Note to print readers; the online version of the Voices column is far more extensive than its pagecount-constrained, paper-based sibling. I think Chano did a good job of pragmatically addressing powerline networking’s strengths and shortcomings, and of doing so a manner that wasn’t overtly DS2-slanted.

Speaking of DS2, shortly after the publication of my August 2 networking hands-on benchmarking cover story and companion online addendums, I was contacted by John Criticos, from the School of Electrical & Information Engineering at the University of the Witwatersrand, Johannesburg, South Africa. John and his partner Brian Berry used Iperf and DS2-based Netgear adapters (along with guidance from both myself and Chano) as the testing foundation for an IEEE paper, ‘Investigating Into Data Communications Using 220 V Mains Within South African Households and Small Businesses‘, to be presented at an upcoming technical conference. The paper’s quite good; I have a final copy of the PDF and will work with John and the conference organizers to provide a download link here at Brian’s Brain post-conference.

John and Brian evaluated, among other things, the powerline networking performance impact of power grid noise sources (notably, they also quantified reflection-induced performance degradation caused by adding variable-length and -impedance open-circuit stubs to the power grid). They tested the well-known interference candidates that my prior writeups had mentioned…televisions, vacuum cleaners, incandescent (but, unfortunately, not flourescent) lamps, heaters, microwave ovens, and hand drills. They also included several potential noise sources that I hadn’t considered; a cell phone charger and laptop and desktop computers, one of which ironically resulted in by far the most significant performance degradation. Below are a few choice quote extracts from the writeup:

The Nokia cell phone charger poses the highest threat to the goodput of the PLC channel resulting in UDP and TCP goodputs of 32.16 Mbps and 43.01 Mbps, respectively [editor note: versus 87.56 and 61.88 Mbps, respectively, in a noise-free environment]…This device caused the SNR value…to decrease by 20.8%. The R&S FS300 spectrum analyzer was used to justify these results by producing a frequency response of the system. It was found that the charger generated noise components with a maximum amplitude of 8 dB, in the 2 MHz to 10 MHz spectrum…All other appliances have negligible effect on the goodput of the channel. To quantify this statement: the heater, which invoked the second highest degradation effect on the channel, yielded a SNR loss of only 1.03%

To prevent overloading the LISN [editor note: line impedance stabilization network], only five appliances were concurrently conencted to the PLC network. Corresponding UDP TCP results are 39.9 Mbps and 29.9 Mbps, respectively…these results indicate a non linear additive degradation effect on the PLC channel.

In both the field and the laboratory, it was found that the Nokia charger yielded the highest channel impairments. This is due to a switch mode power supply (SMPS) integrated in the cell phone charger circuitry. The SMPS actively switches a transistor between full saturation and full cut off. This generates high frequency (HF) noise components that are multiples (harmonics) of the switch frequency. Other makes of cell phone chargers that use a SMPS will impose similar effects onto the PLC channel. SMPS are also integrated into desktop and laptop power supplies; however, these power supplies are equipped with filter circuits that block HF noise. This explains the negligible channel effects imposed by these devices.

A specific combination of appliances resulted in a non-linear degradation effect. This is attributed to the parallel combination of each load’s input impedance. As appliances are added, the effective impedance will decrease. This will alter the degree of multipath fading. The variation in channel degradation may also be a result of the specific length of power cords attached to each load. As shown earlier, the length of a cable segment (power cord) may influence the performance the PLC channel.’

To their last point, I wonder if the added appliances’ incremental interference domination of various potential powerline networking time/frequency ’slots’ is also a factor in the non-linear performance degradation. For as Chano stated in my interview with him:

The interesting feature of noise found in powerline is that it’s not like the famous "white Gaussian noise" found in any digital-communications textbook. It’s "colored" noise (stronger in some frequencies and weaker in others), non-Gaussian (you have very strong peaks that do not follow a normal distribution), and nonstatic (you have shorts periods of silence followed by shorts periods of strong noise). So, a powerline device has to find out which are the "clean" time/frequency slots and make sure to avoid the noisy ones. And once this is done, somebody will plug/unplug something in a room nearby, and you have to repeat the time/frequency analysis all over again, in only a few milliseconds, to ensure that the user does not experience any service interruption. Fortunately, advances in DSP and ASIC technology provide enough computing power to perform a pretty accurate time/frequency analysis of the communication medium, and we are able to ascertain which are the "slots" where we can transmit with efficiencies of up to 10 bits/second/Hz and which are the ones where maximum efficiency is lower (or even zero).

I also wonder if the Apple MacBook power supply plugged in the outlet adjacent to a HomePlug AV adapter in my office is truly, as John and Brian suggest, "equipped with filter circuits that block HF noise"…