Performance Improvement of Bluetooth in the Presence of WLAN DS

Transcription

1 Performance Improvement of Bluetooth Settapong Malisuwan 1, Ph.D. and Thippawan Iamsinthorn 2 1 Department of Electrical Engineering Chulachomklao Royal Military Academy Nakhon-Nayok, Thailand 2 Department of Information Technology Rangsit University Pathumtani, Thailand Astract Bluetooth technology is eing integrated into a variety of personal and moile devices. Bluetooth and Wireless LAN devices use the same 2.4 GHz and and they will interfere each other. In this paper, a systematic analysis is presented to calculate the frame error rate in the Bluetooth systems impaired y electromagnetic interference (EMI) from Wireless LAN transmissions. Considerations of GFSK modulation adopted are duly taken into account. The effects of EM radiation from the Wireless LAN system in the transmission mode on the relevant performance impairment are determined. Simulation studies on Bluetooth communications consistent with the availale details pertinent to EMI from the Wireless LAN system are presented. Furthermore, the orthogonal polarization switching is proposed to overcome the influence of EMI due to unintentional signal from the Wireless LAN system. Relevant conclusions on the reduction of interference y utilizing the polarization-switching technique are discussed. Keywords: EMI, Bluetooth, WLAN, Orthogonal Polarization Switching, Frame Error Rate, and ISM and 1. Introduction This paper considers a IEEE Wireless LAN (WLAN) using a DSSS scheme and having a andwidth equal to 20 MHz. The Wireless Local Area Network (WLAN) system ased on the IEEE standard and the short-range radio ased on Bluetooth share the same ISM and at 2.45 GHz. In an office environment, WLAN and Bluetooth radios will typically operate simultaneously. Because they operate in the same radio and, mutual interference is a concern. The topic of coexistence etween the Bluetooth and WLAN DS has een addressed in several papers. Kammerman [1] reports on tolerale interference levels etween the Bluetooth and devices for various scenarios and device dispositions at various floor locations. Neelakanta [2] reports performance results for the Bluetooth and microwave oven operating simultaneously. However, addressed here is an analysis to evaluate the relevant impairments to the wireless transmissions involved and to deduce the loss in throughput efficiency that an operating Bluetooth system may suffer as a result of EM radiation from the WLAN DS

2 Settapong Malisuwan and Thippawan Iamsinthorn system in the same range. Specifically, the goal of the research presented in this paper is to provide an analytical model for evaluating the coexistence ased on frame error rate (FER). Furthermore, the polarizationswitching strategy is proposed to overcome the influence of EMI due to unintentional signal from the WLAN system. The Bluetooth technology allows any electronic appliance to communicate with other(s) via 2.4 GHz ISM radio-frequency (RF) and over a short range. The operation in these Bluetooth network configurations is ased on frequency hopping technique, which assigns a unique sequence for each piconet. In order to comat, the interference from any neighoring piconet, a frequency hop-selection is adopted. That is, a set of (random) carrier frequencies are selected (in the assigned and), and the transmission of packets are supported on these frequencies, hopped randomly for each transmission. The pattern of frequency hopping is calculated using the address pertinent to the master. (This ensures only one unit eing the master for each piconet, and clocks the transmission events.) There are 10 frequency sequencesfive for the 79-hop system (used in the United States, Europe and most of other countries) and five for the 23-hop system (used in Japan, Spain and France, these countries, however, are expected to adopt the 79-hop system in near future.) The selection scheme consists of two parts: (I) selecting a sequence and (ii) mapping this sequence on the hop frequencies. The prescried hop rate is 1600 hop/second so that the duration spanning each hop is 625 µs. In this time period, the it-data are transmitted with the corresponding sequence of channels adopted eing in pseudorandom from. Both transmitter and receiver use the same pseudo code to tune into the sequence of channel clocked to synchronization. 2. Error Proaility for GFSK In transmission mode, the inary data are fed into a modulator that uses the GFSK scheme. The resulting modulation is centered around some ase frequency. Next, the selected frequency from the hop-selection ox is modulated with the output signal from the GFSK modulator so as to shift the signal to e centered around the selected hop frequency. On the receiver side, the frequency-hopped spread-spectrum signal is pre-demodulated using the same hopfrequency used at the transmitter and then sujected to BFSK demodulation. Hence, the inary data is recovered at the final output. Due to the Bluetooth radio chip uses the GFSK modulation with modulation index etween 0.28 and 0.35 for alancing the effect of in-and interference etween it 0 and it 1 and 0.5 WT factor on gaussian filter. The GFSK is almost similar to FSK modulation. The differences, however, etween the two lie in the implementation considerations. The GFSK modulator is identical to a FSK modulator except that, efore the aseand pulse goes into the FSK modulator, it is passed through a Gaussian filter. This makes the pulse shaped so as to limit its spectral width. Desiraly, this pulse shaping filter should satisfy the following properties: Frequency responses with narrow andwidth and sharp cutoff characteristics, which would suppress the high-frequency components of the transmitted signal; and, an impulse response with relatively low overshoot to avoid excessive deviations in the instantaneous frequency of the FM signal. Hence the response g(t) of this filter is specified y a Gaussian transfer function. In reference to a rectangular pulse of unit amplitude and duration T (centered on the origin), g(t) is given y : [2] 42

3 1 2 g( t) = erfc π WT 2 log2 (1) To find the it error proaility in reference to GFSK scheme, the general form of BER (P c ) developed for FSK [3] can now e modified as follows: ( a + ) / 2 P = Q ( a, ) ( a ) e 0 2 e I (2) As indicated y Proakis [4], with a modulation index equal to 1 and WT, the BER (P c ) corresponding to the traditional FSK is given y: ½ exp [-(E / N o )/2 ]. Now, sustituting WT (or ν - 1), h = 1 (or ρ = 0), a = 0 and = E / N in 0 Eqn (3), the result in that Q (0,) = e 2/2 and I(0) = 1. Therefore, Eqn (2) reduces to P = e e 2 /2 t T (1/ 2) e 1 2 erfc π WT 2 log2 2 /2 t T = (1/ 2) e (3) E / N Frame Error Rate in ACL Packets of Bluetooth TM Transmissions In reference to AUXI packet, there are no FEC and CRC; it means that no error correction and/or detection facilitated. Therefore, the whole packet may fail when any single it error occurs in the packet. For DH1, DH3, DH5, the 16 CRC its are included the packet. But these CRC its enale as error checking only through ARQ scheme, and do not offer error correction. Consequently, oth AUX and DH packets do not have the capaility for error correction. Hence, such packets will e discarded, whenever a single error it occurs. In reference to a DM packet, the information payload plus CRC its are coded with a rate 2/3 FEC, adds 5 parity its to every 10 it segment codeword. This code can correct all the single errors in each codeword. Since the encoder operates with information segments of length 10, tail its with value zero may have to e appended after the CRC it.the total numer of its to e encoded. (that is, payload header, user data, CRC, and tail its) must e a multiple of 10. DM packet 10+5 its Codeword # its Codeword # its Codeword #m Figure 1. DM packet format 43

4 The proaility of error in such DM packets with each code word having the correction capaility to correct the single error, is given y: P c l 15 = 1 k = o k P 15 k e (4) where Pe is the proaility of it error rate, and Pc is proaility of codeword error rate. The corresponding proaility of packet error rate is given y: FER = 1-(1-Pc) m (5) with m eing the numer of codeword in a packet. 4. EM radiation of WLAN versus Bluetooth Transmission: Analysis The unlicensed Industrial, Scientific, and Medical (ISM) frequency and at 2.4 GHz ( GHz GHz) is an attractive and gloally availale alternative for a numer of wireless communication systems. The two main systems utilizing this and are WLAN and the emerging Wireless Personal Area Networks (WPAN). Bluetooth will e the first technology on the market in the WPAN field. WLAN provides wireless data access to a corporate ackone wired network and is ecoming a commonly used solution also for small office and home environments. At the same time, Bluetooth will e integrated into a variety of devices enefiting from the wireless data transfer. Example applications range from desktop and laptop PCs to Personal Digital Assistant (PDA) devices and moile phones and further to wireless headset and wireless sensors. Consequently, the WLAN and luetooth systems will e operating in a close proximity to each other, and often oth k e (1 P ) wireless interfaces are integrated into the same device. Consider the designated time-slots (of 625 µs each) assigned for the Bluetooth transmission from the master-to-the-slave. These slots may accommodate three types of packets, (namely, 1 time-slot packet, 3 time slots packet, or 5 time slots packet). Each packet is set to egin at the instant corresponding to the eginning of even timeslots in the master-to-slave link; and, whenever a packet is sent, it uses the same frequency-hopping channel over its entire duration. The frequency-hopping channel supporting a given packet is specified y the address of the master in the piconet (ULAP: Upper & Lower Address Part) and y the current time- slot position accommodation the packet. This channel is further identified y its designated numer ranging from 0 to 78 (2402 to 2481 MHz), as per the Bluetooth Core Specification V 1.0B. Each of this hopping channel has a andwidth equal to 1 MHz and frequency hopping is set at 1600 hops/s; (hence, the time slot width is equal to 1/1600 = 625 µs as mentioned earlier) Now, the effect of EM radiation from WLAN on the master-to slave Bluetooth communication under discussion can e considered. Figure 2 presents the interference of uplink transmissions from a WLAN station on a Bluetooth device. WLAN Station Bluetooth piconet WLAN Access Point Figure 2. Bluetooth-WLAN coexistence topology 44

5 The measured transmission energy spectrum for the WLAN DS is shown in Figure 3 [5]. The spectrum presents the measured signal spectrum at very close range (elow 0.5m). For a Bluetooth device, this is a predicted to e common operating range. WLAN is designed for longer transmissions periods, ut the Figure 3 presents the interference of uplink transmissions from a WLAN station on a Bluetooth device attached to the same terminal. This WLAN transmission is likely to envelop Bluetooth signals within the same frequencies. The andwidth of the WLAN DS system is roughly 20 MHz. Therefore the proaility that the Bluetooth system will hop into the WLAN DS system is 20/79 or aout 25% [5], [6]. The ensemle of computation carried out refers to simulations ased on the frequency-hopped channels of Bluetooth communication susceptile to EMI as shown in Tale 1: Tale 1. WLAN transmission versus the frequency-hopped channels of Bluetooth communication Energy spectrum of the WLAN DS (GHz) Slot numer of the FH Bluetooth The EM radiation from the WLAN transmission, when occurs, is sufficiently strong enough to cause interference on the Bluetooth transmission so that the entire packet will e dropped if the packet is in its active time and frequency range coinciding the time/ frequency encroachment of the interfering EMI from the WLAN transmission. Figure 3. Transmission energy spectrum measurements of WLAN DS [5] Presently, via simulations performed using the statistics that descrie the occurrence of packet events, details in hopped-frequency channels and random events of coincidence of such channels with the microwave leakage spectrum, the finitepacket loss proaility is estimated. The relevant details are as follows: Details on the transmitted packets - Transmission : Master-to-slave - Numer of packets used : 10,000 - The occurrence of the packets refers to same proaility distriution for all the three types of packets corresponding to 1,3,or 5 time-slots - The time-interval etween successive packets is assumed to e of exponential distriution with a mean value 625 µs (This ensures a Poisson arrival process depicting the statistics occurrence of such packets) Frequency-hopping: - Intially 0x is assigned for the Master address (ULAP) - The clock is set to start at 0x The process is assumed to e in the connection state 45

6 Settapong Malisuwan and Thippawan Iamsinthorn Presented in Tale 2 is computed results performed with the data indicated in Tale 1. The simulations were done with MATLAB. Tale 2 Simulated results in average frame error rate (FER) WLANTX Load Simulation Average FER At this point, there are two FERs can e considered: One due to the packet hopping in the frequency range with a coincidence in time on the EM radiation from the WLAN and the other arising from the considerating relevant to E/No of GFSK. Consequently, the actual FER corresponding to Bluetooth packet interfered y the WLAN is DM3 packet Figure 5. Frame error rate on DM3 Bluetooth packet (1500 its/packet) interfering y WLAN transmission 10-1 *** FER (1.00 WLAN TX Load)... FER (0.75 WLAN TX Load) xxxx FER (0 50 WLAN TX Load) FER actual = FER GFSK FER EMI from WLAN (6) Figures 4-6 show the frame error rate (FER) of 3 types of DM Bluetooth packet with BT = 0.5 and modulation index = 0.32 interfering y WLAN DS transmission with different transmission loads DM5 packet DM1 packet *** FER (1.00 WLAN TX Load)... FER (0.75 WLAN TX Load) xxxx FER (0 50 WLAN TX Load) Figure 6. Frame error rate on DM5 Bluetooth packet (2736 its/packet) interfering y WLAN transmission *** FER (1.00 WLAN TX Load)... FER (0.75 WLAN TX Load) xxxx FER (0 50 WLAN TX Load) Figure 4. Frame error rate on DM1 Bluetooth packet (240 its/packet) interfering y WLAN transmission 46

7 5 Orthogonal Polarization Switching An interesting research that can e pursued comprehensively could e on imposing an orthogonal polarization on each packet transmitted in the frequency-hopping scheme. Considering the two orthogonal polarizations, namely, vertical and horizontal, the random selection of polarization. The polarization switching could reduce the effects of fading and other propagation impairments prevailing in the indoor environments [7]. If the proposed polarization-switching is to e designed, its major circuit (to control the polarization-switching) should e consistent with the Bluetooth aseand/radio specifications. In this case, the transmitter and the receiver units will e hopped in frequency and switched to identical polarizations synchronously. This scheme can e improvised y considering the output of the frequency-hop selection process and extending it through an algorithm to include the polarization-switching (selection). For example, the vertical polarization can e applied whenever odd-numer designated, random frequency channel hopping occurs; and, the horizontal polarization can e applied whenever even-numer designated, randomly hopped, frequency channel is used. A lock diagram of a Bluetooth radio using frequency-hopping superimposed with polarization-switching and GFSK modulation is illustrated in Figure 7. Binary Data MOD BFSK Bandpass filter Spread spectrum signal Gaussian filter Master's Address 79/23 mode Frequency-hopped and polarization-switched carrier output Clock Hop Selection superimposed with Polarization-Switching (a) Spread spectrum signal DEM BFSK Binary Data Master's Address 79/23 mode Frequency-hopped and polarization-switched carrier output Clock Hop Selection superimposed with Polarization-Switching () Figure 7. Bluetooth units with polarization-switching plus frequency hopping of packet transmission capaility. (a) Transmitter and () receiver 47

8 Now, the question is how implement the proposed scheme. It involves two major technological considerations: 1. A hardware requirement of facilitating orthogonal polarizations on the hopped frequency EM waves of the Bluetooth TM transreceivers. 2. Compatile antenna structures on the transreceivers to allow required (switched) polarizations. Both requirements as aove could e essentially implemented y using the relevant concept of diode-switching facilitated on a pair of orthogonal antenna structures [7]. That is, the antennas when switched, could radiate (or receive) vertically or horizontally polarized waves; and such a switching can e randomly done with triggers applied through steering diodes attached to the orthogonal antenna structures. (Alternatively, the transreceive antennas could e configured as circularly-polarized structures, so that they have the capaility to transmit or receive either vertically or horizontally polarized waves). Simulation studies were performed to ascertain the resultant FER in typical Bluetooth systems improvised with the proposed polarization-switching. At this point, there are two FERs can e considered: One due to the packet hopping in the frequency range with a coincidence in time on the EM radiation from the WLAN DS system and the other arising from the considerating relevant to E/N 0 of GFSK. Consequently, the actual FER corresponding to Bluetooth packet interfered y the WLAN is FER actual = FER GFSK FER EMI from WLAN (6) If a rute-forced, polarization-switching is adopted for the Bluetooth transmissions (so as to e diverse from that of microwave oven leakage), the resulting FER will e reduced y a weighting factor (<1) achieved through this polarization diversity. Hence, the resultant FER of Bluetooth transmissions can e specified as FER GFSK FER P where FER P is the polarization-imposed weighting factor *** FER GFSK. FER actual oooo FER improved performance y polarization-switching *** FER GFSK. FER actual oooo FER improved performance y polarization-switching Figure 8. Frame error rate on DM1 Bluetooth packet with BT = 0.5 and modulation index = 0.32 interfering y WLAN Figure 9. Frame error rate on DM3 Bluetooth packet with BT = 0.5 and modulation index = 0.32 interfering y WLAN 48

9 Figures. 8-9 show the frame error rate (FER) versus with different modulation factors. The analysis shown and the simulated results gathered indicate a reduction in the resultant FER of Bluetooth packet transmissions as a result of the proposed polarization-switching. Hence, the scheme advocated here offers a viale strategy to mitigate the invasion of EMI from devices like WLAN on Bluetooth systems. This polarization-switching method can e regarded as etter than the power-control strategy proposed y the authors in [2]. The present technique not only offers no influence on adjacent piconets, ut also takes advantage of polarization-diversity specified performance improvements of indoor wireless transmissions. 6 Conclusions Indicated in this research, the mode of possile interaction etween EMI from the Wireless Local Area Network and Bluetooth transmissions is investigated. Assuming such EMI emerges from ISM and, namely 2.45 GHz and its spectral emission characteristics coincide with the hopped frequencies of frequency hopping transmissions of the Bluetooth communications, the resulting frame error rate (FER) is derived. Considerations of GFSK modulation adopted are duly taken into account. In this research, the orthogonal polarization switching is proposed to overcome the influence of EMI due to unintentional signal from Wireless Local Area Network. In general, the proposed concept can e extended to any CDMA transmissions (either hoppedfrequency or direct-sequence version) and could e adopted for outdoor cellular and/or other indoor technology, such as Wireless Local Area Network. References [1] Kamerman A., Coexistence etween Bluetooth and IEEE CCK solutions to avoid mutual interference, doc.: IEEE /162, Lucent Technologies, July 4, 2000 [2] P.S.Neelakanta, J.Sivaraks and C. Thammakorannonta, Bluetooth-enaled microwave ovens for EMI compatiility, Microwave Journal, Vol. 44, No. 1, , Jan [3] S. Haykin, Communication system 4 th edition, John Wiley & Sons, INC., NewYork, NY, 2000 [4] J.G. Proakis, Digital communication, McGraw Hill, Inc., NY, 1983 [5] M. Hannikainen, T. Rantanen, and J. Ruotsalainen, Coexistence of luetooth and wireless LANs, The proceeding of 8 th ICT 2001, 4-7 June, 2001, Bucharest, Romania [6] J. Zyren, Effect of WBFH power reduction and hop rate, Doc.: IEEE /209- r2, Sept [7] A. Kajiwara, Circular polarization diversity with passive reflects in indoor radio channels, The International Conference on Communications 1997 (ICC 97), Vol. 2, 1997,