Because of the implementation of indoor wireless LAN's and PCN's, it is important to establish a high level simulation model for the indoor wireless communication channel. Hence, collecting the data of bit error rate and frame error rate under different indoor wireless scenarios is mandatory. This paper explores the bursty errors' appearance by measuring BER/FER using a pair of low cost wireless modems. After the data analysis, it seems that there exists some correlation between the bursty errors' appearance and the BER/FER observation.
In recent years, the research of wireless communications receives a lot of attention. From channel measurements and modeling, multiple access techniques, ..., to wireless services and system design, all aspects of wireless communications become the foci of many researchers and communication engineers. Among the various issues of wireless communications, the most fundamental one is the study of the wireless channels. Unlike the line-of-sight radio channels such as satellite channels, wireless channels suffer shadowing and multipath fading phenomena. Therefore, the channels become not only AWGN, but also time-varying and, sometimes, bursty. Hence, the study of wireless channels in different scenarios shows importance.
The purpose of measuring channel's characteristics is to help people understand the link and, furthermore, help people to build up the channel models which are used in all phases of designing a communication system. Measuring indoor wireless channels have been performed by many researchers [1-10]. In [1, 2], the author discussed the propagation channel in light clutter and heavy clutter factories in great detail. He indicated that delay spread values are not correlated with transmitter-receiver distance or topology.  measured the time delay spread in buildings and obtained an rms delay spread value about 110ns. The author also concluded that in his measurements for signalling rates in excess of 400 kbits/s might not feasible for an error probability of 0.001 or less.  did a variety of power-distance relation measurements and obtained the mean path loss exponent n ranging from 1.2 to 6.5 for different situations. In , the authors made a lot of signal level measurements within a steel building that approximates a worst case for building attenuation. In their results, the signal was approximately Rayleigh distributed.  measured the indoor channel at 910MHz and 1.75GHz. The authors concluded that fading was less severe for channels in the 900 MHz band than for channels in the 1.7GHz band. Their result also showed the signal was Rician distributed.  did some measurements (mainly LOS) at 19.37GHz for the future ATM wireless access communication system. The authoers pointed out that with properly oriented high gain antennas, the reduction in rms delay spread and the improvement in K-factor can be achieved.  performed thorough measurements at three different frequencies. Their results showed smaller rms delay spread, higher path loss, and wider coherence bandwidth at higher frequency.  measured the power delay profiles and compared them with ray tracing simulations. Their results showed the ray tracing simulation and the measurement result for LOS were quite matched but for NLOS the simulation and the measurement result were not matched well.  did 850 MHz indoor channel measurements within two dissimilar office building. The author found the results were substantially the same. In , the authors studied the UHF propagation in modern office buildings. They concluded that propagation through the clear space between furnishings and ceiling results in excess path loss for distances beyond that at which the Fresnel zone fills the clear zone. They also concluded that ray tracing accounts only for specular reflection and transmission.  measured the delay spreads and signal level at 850MHz at residences and medium sized office building. The author summarized that for the worst-case delay spread of 325ns, it can supported up to 250kbps for 0.001 error probability, while for the worst-case delay spread of 100ns, it can supported up to 800kbps for 0.001 error probability. [13, 14] models the path losses in a building and multifloored buildings, respectively.  used only a simple 10nlog(d) model, while  founded the value n for short distance is smaller than that for long distance. The reason was explained in .
The aforementioned works emphasized the following: measuring the path loss [1, 4, 8, 13, 14], recording the power-delay profile and analyzing delay spread statistics [2, 3, 6, 7, 8, 9, 10, 12], relating the measured data with the prediction by field theory or ray tracing [9, 11], and fitting the stochastic model to the measured data [1, 5, 6, 7, 8]. The BER analysis was only minor works of some researchers [3, 8, 12]. This motivates us to measure the BER/FER of indoor wireless channels and observe the bursty errors' appearance. Using the results we obtained, we are still trying to establish a more abstract indoor channel model.
In this project, we build up a pair of low cost 64Kbps wireless modems for 902~928 MHz ISM band. The cost of each modem is less than NT$500. Using these low cost modems, we measured a viariety of environments including: open space, hallway, laboratories, ..., and factories. This rest of this paper is organized as follows. In Section 2, the measurement system is described. In Section 3, some measurement results for several different situations are shown, Finally in Section 4, a summary of this work is given.
The block diagram of the wireless modem is dicpited in Figure 1. The SST ASIC AIC 9001 and the RF module ARF 9003 are products of the ALFA, Inc., Taiwan. The AIC 9001 is a DSSS IC with chip length 32 and maximum data rate 160kbps (half duplex). The ARF 9003, the corresponding RF module of AIC 9001, operates at frequency range in 902~928MHz. It provides 70 overlapped channels or 10 non-overlapped channels. The modulation scheme of the ARF 9003 is GMSK. The measured output power is 16dBm at 5V or 14 dBm at 4.5V power supply.
The 89C51 micro processor is the core of the modem. It controls the data flow and protocols. It also connects to the outside world (the PC in Figure 1 or other RS-232 compatible device) via its UART port. Because the clock speed of the 89C51 is low (about 16MHz), the speed of the whole system is limited by the 89C51. As we previously mentioned, the data rate of our modem is only 64kbps.
The channel measurement experiment setup is shown in Figure 2. Two wireless modems are connected to two notebook PC's as the tranmitter and the receiver. 60 frames are sent from the trasmitter to the receiver in each experiment. Each frame contains 1944 bits. All data measured are recorded in the receiver notebook PC for later analysis. The equipment was first tested on the top of a hill for the line of sight test. There were no obstructions at all and the view is clear. The BER is under as the transmitter and the receiver are located 5 up to 60 meters apart. The experiments were then performed in a variety of indoor environments.
Fading of a fixed indoor wireless channel mainly comes from the movement of personnel. Herein, we show some measurement results:
The BER we obtained is less than as the transmitter and the receiver are located 5 up to 20 meters apart. No burst-type error is founded in this situation.
The BER/FER results are given in Table 1. We observed short bursty error when the T-R distance is greater than 30 meters. This is believed as the influence of multimath propagations.
In the same factory as was mentioned in case A, we repeat the experiments in the non-LOS situations. The results are shortly summarized in Table 2. On can see that the shadowing effect is very obvious. Figure 3 shows the histogram of frames in which some bursty errors occur for longer T-R distance. This means the channel is no longer AWGN. Among the data we collected, we found one frame with 50 erring bits in 15m data and one frame with 66 erring bits in 20m data (see Figure 4). These two data are dramatically greater than other data of their groups. The corresponding bursts are also largely greater than other bursts. Figure 5 shows scattering plots of errings bits vs. bursts. The slope of the linear regression line is getting larger for larger T-R distance. This is due to the larger bursty length and bursty rate for larger T-R distance. The average bursty length for T-R distance of 10m, 15m, and 20m is shown in Figure 6. Figure 7 is the scattering plots of erring bits vs. average bursty length of each frame for different T-R distances. The slopes of the regression lines in Figure 7 are closer than those in Figure 5. Finally Fighre 8 shows the percentage of bursty errors out of the total errors for four different T-R distances.
In this work, we build up a pair of low cost wireless modems first. Using these modems, we measured the BER/FER and recorded the bursty errors of a viariety of fixed indoor wireless channels. Among our measurements, we found that the quality of transmission of sending message from a room to the receiver outside the room is worst. However, it is still too early to draw solid conclusions here. More measurements and analysis are needed.
 T. S. Rappaport, "Indoor Radio Communication for Factories of the Future," IEEE Commun. Mag., vol. 27, no. 5, May 1989.
 T. S. Rappaport, "Characterization of UHF Multipath Radio Channels in Factory Buildings," IEEE Trans. Antennas Propagat., vol. AP-37, no. 8, Aug. 1989.
 D. M. J. Devasirvatham, "Time Delay Spread Measurements of Wideband Radio Signals within a Building," Electron. Lett., vol. 20, no. 23, Nov. 1984.
 S. E. Alexander, "Characterising Buildigs for Propagation at 900 MHz," Electron. Lett., vol. 19, no. 20, Sep. 1983.
 H. H. Hoffman and D. C. Cox, "Attenuation of 900 MHz Radio Waves Propagating into a Metal Building," IEEE Trans. Antennas Propagat., vol. AP-30, no. 4, July 1982.
 R. J. C. Bultitude, S. A. Mahmoud, and W. A. Sullivan, "A Comparison of Indoor Radio Propagation Characteristics at 910MHz and 1.75GHz," IEEE J. Selected Areas Commun., vol. 7, no. 1, Jan. 1989.
 Y. Sun, et al., "Indoor Channel Characterization Measurements with Directional Antennas for Future High Frequency ATM Wireless Access Systems," Proc. IEEE Int'l Symp. Personal, Indoor, and Mobile Radio Commun., PIMRC, vol. 1, 1997.
 G. J. M. Janssen, P. A. Stigter, and R. Prasad, "Wideband Indoor Channel Measurements and BER Analysis of Frequency Selective Multipath Channels at 2.4, 4.75, and 11.5 GHz," IEEE Trans. Commun., vol. 44, no. 10, Oct. 1996.
 K. W. Cheung and R. D. Murch, "Measurement, Characterization, and Modeling of the Wideband ndoor Channel," Proc. IEEE Veh. Technol. Conf., vol. 1, 1996.
 D. M. J. Devasirvatham, "A Comparison of Time Delay Spread and Signal Level Measurements Within Two Dissimilar Office Buildings," IEEE Trans. Antennas Propagat., vol. AP-35, no. 3, Mar. 1987.
 W. Honcharenko, et al., "Machanisms Governing UHF Propagation on Single Floors in Modern Office Buildings," IEEE Trans. Veh. Technol., vol. 41, no. 11, Nov. 1992.
 D. M. J. Devasirvatham, "Time Delay Spread and Signal Level Measurements of 850 MHz Radio Waves in Building Environments," IEEE Trans. Antennas Propagat., vol. AP-34, no. 11, Nov. 1986.
 J.-F. Lafortune and M. Lecours, "Measurement and Modeling of Propagation Losses in a Building at 900MHz," IEEE Trans. Veh. Technol., vol. 39, no. 2, May 1990.
 S. Y. Seidel and T. S. Rappaport, "914 MHz Path Loss Prediction Models for Indoor Wireless Communications in Multifloored Buildings," IEEE Trans. Antennas Propagat., vol. 40, no. 2, Feb. 1992.