 |
| EMagazine Volume 5 Issue 4 Aug / Sep 2000 |
Features |
|
|
|
Interviews |
|
Free
Email Updates |
|
|
 |
| Home
Toys Article - August 2000 - [HTI Home Page] |
Read All About It News Announcements - Updated Hourly [Click Message To Learn More] |
|
ARE RELIABLE
POWERLINE COMMUNICATIONS POSSIBLE? We believe that PLC transceiver manufacturers will benefit from adding FEC (Forward Error Correction) to their designs, and we would like to suggest that standard bodies (EIA, ANSI, etc.) automatically add FEC coding into their future PL communication standards. We would like to suggest that the EIA revise the EIA 600 (CEBus) and EIA 709 (associated to LonWorks) standards to include at least a minimal FEC coding algorithm. Reliable powerline communications are possible with FEC coding! |
Metricom Corp. designed the PLC-1, a high performance PLC transceiver IC comprising a high efficiency Forward Error Correction technique. For more information, see http://www.metricom-corp.com |
There is a growing interest in the use of PowerLine Carrier (PLC) for data communication using the intrabuilding electric power distribution circuits. Power lines were not designed for data communications and exhibit highly variable levels of impedance, signal attenuation and noise. Many studies are available that describe in some detail the impedance, signal attenuation and noise characteristics of powerline networks. In their study of residential powerline noise sources, Vines et al. [1] identified 4 types:
- Sources that generate impulse noise in synchronism with the 60 Hz power frequency.
- Smooth spectrum noise generated by loads not synchronous with the power frequency (e.g., the universal motor in an electric drill).
- Non-synchronous, single-event impulse noise, (e.g., thermostat or light switching).
- Non-synchronous periodic noise.
In general, intrabuilding powerline noise consists of continuous, relatively
low-level background noise punctuated by high-level noise impulses. Background
noise is typically Gaussian [2], and its effects on communication performance
are well understood [3]. In the case of impulse noise, its time-domain
characteristics (amplitude, width and interarrival time) are very important to
determine the influence on data communication systems.
Chan and Donaldson [4] characterized noise impulses on PLC networks. They concluded the following:
- Impulse strength is typically more than 10 dB above the background noise level and can exceed 40 dB.
- Impulse frequency for the dominant impulse train is usually 120 Hz.
- Impulse width can vary up to a few percentage points of the impulse period for 120 Hz impulse noise.
- Because both the noise and the wanted signal are subject to attenuation, noise sources close to the receiver will have the greatest effect on the received noise structure, especially when network attenuation is substantial.
- Harmful effects of impulse noise on data communication systems are expected.
Item 4 above bears significant consequences as it implies that when a noise source is located close to a receiver and when the signal is attenuated (across-phase communication, or attenuation caused by the combination of line impedance and the presence of low impedance loads along the communication link) a local noise source could make a receiver exceed its noise tolerance (signal to noise ratio), yielding erroneous data on the receiver end.
The single-event noise type can be easily overcome by repeating the data packet. However, the situation is more complex with the line-synchronized noise type, because this type of noise can introduce one error (or more) 120 times per second. If the data transmission process lasts more than 1/120 second (or 8.33 ms), errors will likely occur, and packets will be lost. Repeating the packet will likely be unable to circumvent the problem in many circumstances. Breaking the data packet into smaller segments lasting less than 8.33 ms, and transmitting these segments between noise bursts is also impossible, because the transmitter could have a local noise source, different from the noise source at the receiver. In other words (and remembering item 4 above), in any PLC system, one should assume that noise at the receiver is unknown to the transmitter. Consequently, it could be difficult (if not impractical) for an intelligent transmitter to analyse the powerline noise characteristics and adapt its transmission strategy, because noise at the receiver end is very likely to be different from noise measured at the transmitter. Moreover, such an intelligent transmitter could not broadcast messages to N receivers, each receiver having its own and different local noise characteristics.
Local noise is not the sole source of communication errors. Sudden impedance variation can also induce similar effects. Different loads (some compact fluorescents, personal computer power supplies, switching power supplies) are known to make the impedance of the communication link change abruptly with time, generally synchronously with the powerline frequency. (These loads generally offer a higher impedance level around the zero crossing of the power wave.) For the same reasons as above, impedance variations at the receiver end are almost undetectable at the transmitter location, making adaptive algorithms difficult to implement. The same principles apply to impedance variation: more severe effects are generally observed when the impairment source is close to the receiver, and the impedance between the receiver and the transmitter is high. Once again, repeating a longer than 8.33 ms packet will likely not overcome line synchronous, impedance variation related error bursts.
POWERLINE COMMUNICATION ERRORS EXIST!
When analysing the different PLC technologies available on the market today, one notices two fundamental and distinct design approaches. Certain well-known PLC technologies assume that the modulation and demodulation technique they implement is sufficient to guarantee the minimal threshold of reliability that the consumer demands. Such technologies would have us believe that they possess an ultra-sturdy Magic Bullet of sorts that will lay waste to the obstacles on its path. Such technologies give little importance to communication error management and do not implement Forward Error Correction (FEC).
The second, and in our opinion, more prudent and more realistic approach assumes that a PL communication channel is an extremely difficult medium and that communication errors will occur. Technologies conceived from that point of view actively manage communication errors and implement, for example, an FEC technique.
Granted, the Magic Bullet approach must also be implemented when designing a PLC technology, in that the chosen modulation/demodulation technique must be of the utmost sturdiness. For example, older PL techniques used an On-Off-Keying type amplitude modulation in which a silence, i.e., the absence of a carrier, represents the zero symbol. This type of modulation is, of course, sensitive to all noise sources in its band, even noise of low amplitude, since the presence of any noise changes a zero symbol (0) to a one symbol (1). It is now widely known that FSK modulation, for example, offers superior sturdiness.
Having chosen the modulation technique best suited to the communication medium, elementary caution and field-testing will then demonstrate that there are cases where communication errors exist and persist, despite the Magic Bullet. Unless consumers are willing to accept low reliability-an improbable position on the part of consumers-communication error management is, in our opinion, an absolute necessity, and all the more so considering that the PL medium is among the most challenging.
Most Powerline Carrier systems and technologies offer some kind of error detection technique, (like Cyclic Redundancy Check or CRC) so that erroneous packets are not interpreted. For example, the CEBus standard includes an excellent CRC-based error detection algorithm for the PowerLine medium. Further enhancing the possibility of reliable communications in hostile environments requires appropriate signalling schemes and error control strategies. It is well known that FEC techniques [5] can significantly improve communication reliability.
ERROR MANAGEMENT
Simply stated, FEC adds redundant information to the original message, allowing the receiver to retrieve the message even if it contains erroneous bits. Nowadays, FEC is everywhere around us; it is extensively used in many (if not most) digital communication systems, in Compact Disk technology, (to tolerate scratched CDs), in satellite communications, etc. Unfortunately, only a few PLC technologies offer both an efficient Forward Error Correction and an error detection algorithm. Facing line synchronous error bursts, (and particularly when item 4 above is involved) typical non-FEC systems are often unable to deliver any usable data. They only detect that the message has one or more errors, but are unable to correct the erroneous bit(s) in the received packet, and simply reject the packet even if a single bit (out of hundreds bits) is incorrect.
On intrabuilding Powerline networks, Donaldson et al [6] observed typical FEC coding gains of 15 dB at 10 -3 decoded Bit Error Rate (BER). (Our own experiments also show spectacular results.) This is a very significant improvement in noise tolerance, (Signal-to-noise- ratio) as well as other impairments tolerance like sudden impedance variation. " Stated differently, FEC coding typically enhances decoded Bit Error Rate values three order of magnitude " relative to systems not implementing Forward Error Correction techniques." [6]
FEC is a well-known and relatively simple technique that significantly increases communication reliability in noisy channels like Powerline. In PLC communication, FEC enhanced products should also mean much less unsatisfied customer call-back, less installation and debugging time, less (if any) filters to install in front of noisy loads, etc. FEC could mean thousands and thousands of Dollars savings per year to PLC manufacturers/dealers/installers. FEC has the potential to dramatically increase customer satisfaction, translating into increased sales.
But FEC is not totally "free "! It involves adding redundant data, meaning that more bits per message must be carried out. In other words, it decreases the throughput (number of effective data bits delivered per second) in noiseless environments, where FEC would not be mandatory. But powerline is not such a noiseless environment. In noisy networks, FEC can even increase the net throughput over non-FEC systems, by reducing the need for packet repeating. In severe situations, (particularly in presence of line synchronous impairments, causing error bursts) FEC enabled products can often deliver data where other non-FEC products simply crash. On the other hand, implementing a FEC algorithm in a powerline modem IC is likely to be free: FEC implementation is relatively simple and should not require a significant increase of the die size of a transceiver or modem IC.
Conclusion
Regrettably, many manufacturers and consumers have good reason to believe that PL communications are unreliable. Indeed, older and less sophisticated technologies-as well as more recent ones that do not implement an adequate FEC technique-contribute to a large extent in the spreading of the belief that PL communications are synonymous with unreliability.
Yet the PL medium is of paramount interest and does allow for sufficient reliability to meet application requirements, as long as one keeps in mind that this medium is among the most challenging and thus demands the design of communication technologies suited to the task at hand, therefore including an FEC technique.
- Given the very significant reliability increase obtained from FEC,
- considering that a power line is probably the most error-prone communication medium,
- given the fact that cost of FEC is null when added into a PLC transceiver IC,
- considering the benefits and savings that FEC brings to customers, installers, distributors and manufacturers of PLC products,
- given the economic losses potentially incurred with non-FEC systems,
we believe that it is not advisable, nor affordable for anyone to offer Powerline Carrier products that do not implement an efficient Forward Error Correction algorithm. From an economic and marketing perspective, FEC is not only free: it can save lots of money; it increases customer satisfaction, helping generate more sales.
We believe that PLC transceiver manufacturers will benefit from adding FEC to their designs, and we would like to suggest that standard bodies (EIA, ANSI, etc.) automatically add FEC coding into their future PL communication standards. We would like to suggest that the EIA revise the EIA 600 (CEBus) and EIA 709 (associated to LonWorks) standards to include at least a minimal FEC coding algorithm.
Reliable powerline communications are possible with FEC coding!
[1] R.M. Vines, M.J. Trussel, L.J. Gales and J.B. O'Neal, Jr., " Noise on residential power distribution circuits, " IEEE Trans. Electromagn. Compat., Vol. EMC-26, pp.161-168, Nov. 1984
[2] H.J. Trussel and J.D. Wang, " The effect of hard limiters on signal detection in harmonic noise using adaptive noise cancellation, " IEEE Trans. Pwr. Del., Vol. PWRD-1, pp. 73-78, Jan 1986.
[3] J.M. Wozencraft and I.M. Jacobs, " Principles of communication Engineering, " New York: Wiley, 1965.
[4] M.H.L. Chan and R.W. Donaldson, " Amplitude, width and interarrival distributions for noise impulses on intra-building powerline communication networks, " IEEE Trans. Electromagn. Compat. Vol. EMC-31, pp320-323, Aug 1989
[5] S. Lin and D.J. Costello, Jr., " Error Control Coding: Fundamentals and applications. " Englewood Cliffs, N.J.: Prentice-Hall, 1983
[6] M.H.L. Chan, D. Friedman and R.W " Donaldson, " Performance enhancement using forward error correction on power line communication channels. " SM 93 367-3 PWRD presented at the IEEE/PES Summer Meeting, Vancouver B.C., July 1993
