Wireless LANs Title: Effects of Microwave Interference On IEEE 802.11 WLAN Reliability

The influence of microwave oven interference on IEEE802.11 Wireless Local Area Network (WLAN) performance is a significant factor because they share common spectrum in the 2.400 2.4835 GHz Industry, Science, and Medicine (ISM) band. FCC regulations permit radiated power of up to 1 watt in this band provided spread spectrum techniques are employed. Spread spectrum methods facilitate multiple users sharing the same spectrum in an unlicensed environment and offer interference rejection properties. There are two spread spectrum techniques addressed by FCC regulations (15.247). These are Direct Sequence Spread Spectrum (DSSS) and Frequency Hopped Spread Spectrum (FHSS). Because of the significant differences in the two methods, the effects of microwave oven (MWO) interference are quite different on systems employing these techniques. This paper describes MWO interference and presents a model, which is useful in predicting WLAN reliability. The mechanisms by which the interference disrupts system performance for DSSS and FHSS are described separately. Finally, quantitative results showing packet error rate (PER) under varying levels of interference and packet length are presented and discussed. May 1998 doc: IEEE P802.11−98/240 Submission Page 2 Intersil Corporation Summary This paper describes the results of an analysis of the effects of varying packet length and interference level on the reliability of WLANs in the presence of microwave oven (MWO) interference. There are four different aspects of this analysis. Section I deals with modeling interference from microwave ovens. The results of an NTIA report on interference from MWO in the 2.4 GHz ISM band are summarized along with some relevant journal articles. A model of MWO interference presented by Motorola before the IEEE 802.11 WLAN Working Group is discussed. A MWO can be effectively modeled as a swept narrowband jammer with a 50% duty cycle. The resulting interference is synchronized to the 60 Hz AC power line voltage due to the fact that the magnetron power supplies are only half wave rectified. Section II includes a basic review of the performance of FHSS systems in the presence of narrow band jammers. FHSS systems combat MWO interference by avoiding it. Performance curves are presented which show Packet Error Rate (PER) as a function of both packet length and interference level. Based on the model presented, it is shown that the best line of defense for an FHSS system is a short packet length. This will permit the successful transmission of smaller packets between bursts of interference. Section III extends this discussion to DSSS systems. DSSS systems have wide occupied bandwidths. This increases the probability that MWO interference will fall “in band”. However, the effect of processing gain and the underlying modulation method must be considered. The DBPSK/DQPSK modulation method employed in DSSS radios is considerably more robust than the 2FSK/4FSK method employed by IEEE 802.11 FHSS systems. In addition, the despreading process spreads the bulk jammer power out of band, giving an additional 10 dB improvement in radio performance over non-spread methods. The remaining in-band noise is incoherent white noise. DSSS systems deal with MWO interference by suppressing it, not by avoiding it. Section IV summarizes the data and provides an interpretation. The results demonstrate that FHSS receivers can transmit short packets (100 200 bytes) even in even a very noisy environment. However, when using longer packets (1000 bytes), FHSS systems require a signal strength of 16 to 17 dB above peak interference levels to achieve reliable operation in the presence of MWO interference when operating at 1 Mbps. This effect is even more pronounced when operating at 2 Mbps. By contrast, packet error rates can be high even for short packets when interference levels exceed signal strength in DSSS systems. Once signal power is roughly equal to jammer power, the DSSS systems can provide reliable operation, regardless of packet length. A DSSS system can reliably receive 1000 byte packets with a signal-to-jammer power ratio of roughly 0 dB, based on the analysis presented. Experience has shown that DSSS systems can operate reliably even in very close proximity to a microwave oven. Analysis such as this provide a good framework for discussion of the MWO interference question, but the most convincing method is a side-by-side test of FHSS and DSSS systems in the presence of an operating microwave oven. May 1998 doc: IEEE P802.11−98/240 Submission Page 3 Intersil Corporation Section I: Model of Microwave Oven Interference The results of extensive measurements of interference form fourteen different microwave ovens were summarized in a two volume NTIA report [1,2]. The report demonstrates that the interference has roughly a 50% duty cycle with a 16.7 msec period. This is due to the fact that the magnetrons in the ovens are driven by 60 Hz AC power and are active during only half of the sinusoidal line voltage cycle. Frequency domain measurement of one of the tested MWO is shown in Figure I-1. The measurements were taken using a spectrum analyzer in the “max hold” mode. Peak levels and interference bandwidth vary considerably among ovens from different manufacturers. Figure I-1 “Max-Hold” Amplitude vs. Frequency Plot of MWO Interference from NTIA report The results of the NTIA tests are informative, but the “max hold” frequency domain measurements give an overly pessimistic view of the MWO interference problem. Subsequent measurements [3] using spectrographic techniques show that the instantaneous interference is actually very narrow band. Further, the frequency is swept over a significant portion of the ISM band as the power line voltage across the magnetron varies on each positive half cycle of the 60 Hz sinusoid. A representation of a spectrographic plot is shown in Figure I-2. Of critical importance are the swept frequency range (fswept) and channel dwell time (tdwell). It has been pointed out that any time an FHSS radio dwells on a channel within the range of swept frequencies for more than 16.7 msec, it will be hit at least twice by interference [4]. As previously mentioned, fsweep varies considerably among ovens from different manufacturers. May 1998 doc: IEEE P802.11−98/240 Submission Page 4 Intersil Corporation Occupied Channel