Bluetooth and WLAN

Introduction
Bluetooth Radio interface
WLAN Radio interface
Mutual Interference
  - Bluetooth interferes WLAN
  - WLAN interferes Bluetooth
  - Measurements
Co-existance of Bluetooth and WLAN
Comparison of different access technologies

Introduction

This chapter will first provide an overview over the radio interfaces of Bluetooth and WLAN (IEEE 802.11), as they are essential for the understanding of the discussion "co-existance, competition and interference".
For the radio transmission Bluetooth and WLAN (IEEE 802.11b) both uses the unprotected ISM-band (Industrial, Scientific, Medical) at 2.4 GHz. In the US and in most countries of Europe, a band of 83.5 MHz width is available. To understand the issue of coexistence between Bluetooth and WLAN a short description of both radio types is necessary.
The final part of this chapter will then address why the project team supports the vision of co-existance between the two standards.

Bluetooth Radio Interface

For Bluetooth transmission 79 RF channels spaced 1 MHz apart are defined. The transmitting frequencies in the majority of countries can be calculated by:
f = (2402 + k) MHz                     k= 0,...,78 (channel number) 
For the transfer of the digital data symbols the signal is modulated with the help of Gaussian Frequency Shift Keying (GFSK). Thereby the transmitting frequency for the transfer of a logical ONE is increased by at least 115 kHz and is correspondingly decreased for the transfer of a logical ZERO. To avoid a widening of the power density spectrum and therefore to avoid a high power consumption, the steep transitions are rounded off in the base band by the use of a Gaussian Filter. GFSK distinguishes itself by a cheap and simple modulation technique. The channel bandwidth (20dB transmit bandwidth) is about 1 MHz and a gross transmission rate of about one 1 Mbit/s is reached.

In the standard the transmitting power level is classified into three classes: 0 dBm = 1mW, 4 dBm = 2,5 mW and 20 dBm = 100 mW. Products with a transmitting power of 4 dBm are currently only poor represented on the market. With 1 mW transmitting power a range of approximately 10 m can be reached and with 100 mW distances of 30 m to 100 m are achieved. The requirement for a Bluetooth receiver is an actual sensitivity level of -70 dBm or better. To increase the range at small transmitting power, products with -90 dBm receiving sensitivity are announced.

A Frequency Hopping Spread Spectrum technology (FHSS) is applied to combat interference and fading. The Bluetooth channel is represented by a pseudo-random hopping sequence hopping through the 79 RF channels. The hopping sequence is unique for the generated net and is determined by the Bluetooth device address of the master. The phase in the hopping sequence is determined by the Bluetooth clock of the master. The channel is divided into time slots where each slot corresponds to an RF hop. Consecutive hops correspond to different RF hop frequencies. The nominal hop rate is 1600 hops/s and therefore the slot length is 625 ms. All Bluetooth units participating in one net are time and hop synchronised to the channel. Prior to participating in a net the initiating Bluetooth node most enter the paging mode. Paging requires an average of 1.3 s to complete. During paging, the involved devices hop at 3200 hops/s and transmit very short ID packets with a duration of only 70 msec.With frequency hopping it is possible to get statistically out of the way of other interferers in the ISM-band. On the other side Bluetooth is now acting himself as interferer distributed in time over the whole ISM-Band. Signal break downs due to interferences caused by multiway spreading are limited to the short hopping periods. A disadvantage caused by frequency hopping is the long time of up to 5 s (sometimes 10 s) to build up the Bluetooth connections.

For fullduplex transmission, a Time-Division Duplex (TDD) scheme is used. On the channel, information is exchanged through packets. Each packet is transmitted on a different hop frequency. A packet nominally covers a single slot, but can be extended to cover up to five slots. The Bluetooth protocol uses a combination of circuit and packet switching. Slots can be reserved for synchronous packets. Bluetooth can support an asynchronous data channel, up to three simultaneous synchronous voice channels, or a channel which simultaneously supports asynchronous data and synchronous voice. Each voice channel supports a 64 kb/s synchronous (voice) channel in each direction. The asynchronous channel can support maximal 723.2 kb/s asymmetric (and still up to 57.6 kb/s in the return direction), or 433.9 kb/s symmetric. These data rates are reduced by other systems in the ISM band like microwave ovens, WLAN, HomeRF, door opening systems or other independent Bluetooth nets. 

WLAN Radio Interface

For 802.11b systems both Frequency Hopping Spread Spectrum(FHSS) and Direct Sequence Spread Spectrum (DSSS)technologies are defined in the IEEE standard. However almost all of the 802.11b products don’t supply FHSS and therefore the following is restricted to DSSS.
A DSSS system spreads the baseband data by directly multiplyingthe baseband data pulses with a pseudo-noise sequence (PN) that is produced by a pseudo-noise code generator. A single pulse or a symbol of the PN waveform is called a chip. One data bit is therefore expressed by several chips and this "spreads" the data into a large coded stream that takes the full bandwidth of the channel. Interferences are expected to emerge only in a small frequency part of the channel. At the receiver, multiplication with the spreading waveform generates the data signal with its small bandwidth and smuts the interfering signal over the whole channel bandwidth. In 802.11b systems the channel bandwidth is 22 MHz and a chipping rate of 11 MHz is used. As PN sequence a 11-chip Barker code or 8-chip Complementary Code Keying (CCK) is used. For modulation Differential Quadrature Phase Shift Keying (DQPSK)or Differential Binary Phase Shift Keying (DBPSK) is applied.As shown in table 1 this results in bit rates up to 11 Mbit/s with fallback modes of 5,5 Mbit/s, 2 Mbit/s and 1 Mbit/s.
Table 1: Key characteristics of IEEE 802.11

 
Bit Rate Spreading Modulation Symbol Rate
11 Mbit/s CCK DQPSK 1.375 MSps
5.5 Mbit/s CCK DQPSK 1.375 MSps
2 Mbit/s Barker DQPSK 1.0 MSps
1 Mbit/s Barker DBPSK 1.0 MSps

Typical transfer rates for user data are 5 Mbit/s. For difficult propagation conditions (i.e. larger range, interference, ...), the system uses link adaptation to lower transfer rates. The next table gives an overview about the different data rates on the physical layer of 802.11b systems and the corresponding maximum range for open environments (i.e., outdoor or large halls) and for “closed” environments (i.e. indoor):

Table 2: Use cases (range and capacity) for IEEE 802.11
 

11 Mbit/s
5.5 Mbit/s
2 Mbit/s
1 Mbit/s
Open environment; range up to
150 m
250 m
300 m
400 m
Closes Environment, range up to
30 m
35 m
40 m
50 m

Note that the figures for the ranges are average values which may vary significantly depending on the specific environment structure.

According to European regulations, in the ISM band, ranging from 2.400 to 2.4835 GHz, there are 13 overlapping channels with a separation of 5 MHzavailable for WLANs (with a very few exceptions in some countries). Avoiding interference the minimum distance between the centre frequencies is at least 25 MHz. Therefore, up to three non-overlapping channels are available in the ISM band. Studies show that adjacent cells will not interfere with each other when the channel spacing uses channel center frequencies that are 15 MHz apart. The transmitted power of WLAN systems usually is 100 mW. At a data rate of 11 Mbit/s the receiver sensitivity should be at least-76 dBm.

Additional information concerning WLAN and other 802.11 systems, DECT and Home RF in comparison to Bluetooth is given in the project report of PIR 3.6: „State-of-the-art-study: Alternative technologies and mechanisms“ (P1118-doc36.pdf ).
 

Fig. 1: WLAN and Bluetooth RF channels in the ISM-Band at 2.4 GHz in most European countries

Mutual interferences

Fig. 1 shows some possible WLAN channels and the Bluetooth RF channels available in most European countries. Bluetooth is hopping over 79 RF channels with a bandwidth of 1 MHz. It can be estimated that a frequency hop of one active Bluetooth transmitter overlaps a WLAN channel with a probability of about 20% - 25%since the power density at the boarder of a WLAN channel decreases.
In order to determine the degree to which the radios will cause harmful interference to each other, a number of assumptions are necessary. It is difficult, if not impossible, to define a “typical” network topology. User scenarios and even indoor propagation models can be rather subjective. However, by using some reasonable assumptions, analysis of the interference caused by co-location of the two radio types can proceed.
These assumptions must include:
a. a network topology and user density
b. propagation model
c. network traffic loads for IEEE 802.11b and Bluetooth
A simplified indoor propagation model has been proposed by A.Kamerman, Lucent Technologies. Line-of-sight propagation is assumed for the first 8 meters. Beyond this point, path loss increases as a function of rn, where r is range and n = 3.3. This can be expressed in terms of decibels:
Lpath=20 log (4 pi  r / lambda ),                     r < 8 m      (1)
        = 58.3 + 33 log ( r / 8 ),                        r > 8 m      (2)
where:
    lambda  = free space wavelength @ 2.45 GHz (0.1224 m)
    r = range (m)

Bluetooth interferes WLAN

The impact of Bluetooth personal area networks on a WLAN system is investigated in a paper written by Jim Zyren with the title: „The reliability of IEEE 802.11 Hi Rate DSSS WLANs in a high density Bluetooth environment“ (Bluetooth-WLAN-interference.pdf).
A high density environment has been postulated. Large numbers of both types of devices are present within the topology analysed. In addition, traffic loads are assumed for the Bluetooth piconets. This paper focuses exclusively on the reliability of the IEEE 802.11b high speed wireless network in the presence of interference from Bluetooth radios.
For the purpose of this analysis, the following assumptions were made:
· WLAN mobile station may be located up to 20 meters from theWLANaccess point
· The average density is one WLAN mobile station every 25 sq. meters
· The transmitter power for both WLAN mobile nodes and the WLAN access point is +20dBm
· There is one Bluetooth piconet co-located with each WLAN node.
· The Bluetooth piconet consists of two or more Bluetooth devices which are capable of establishing at least a point-to-point link.
· The Bluetooth devices are limited to 0 dBm transmit power.
The degree to which an IEEE 802.11b terminal is susceptible to interference from nearby Bluetooth transmitters is clearly dependent upon the strength of the desired DSSS signal from the access point.

An 11 Mbps DSSS radio can provide reliable service with a narrow band interferer (such as a Bluetooth transmitter) falling within its pass band as long as the Signal-to-Interference Ratio (SIR) is greater than roughly 10 dB. This approximation is conservative and has been verified in lab tests. Therefore, if the Bluetooth signal is more than 10 dB below the DSSS signal, it will not cause significant interference. However, when the Bluetooth interference exceeds the 10 dB SIR threshold, the DSSS terminal will experience a dropped packet provided there is an overlap in time and frequency. Therefore, the number of potential Bluetooth interferers to which a DSSS node is exposed to depends on the range from the AP.

However, there is only about a 25% probability that an active Bluetooth transmitter will be in the DSSS passband on any given hop period. It should also be mentioned that the probability of collision is further reduced by the fact that the Bluetooth transmitter is only active for 366 msec in each 625 msec Bluetooth hopping period.

Accurate estimates of IEEE 802.11b network throughput in the presence of a single but fully loaded Bluetooth interferer must account for the possibility of Bluetooth collisions with ACK packets, and network overhead associated with re-contention for the network in the event the transmitting station fails to receive an ACK. This results in a drop down of the WLAN throughput rate from about 6 Mbit/s without interferences to 3.5 Mbit/s at a typical packet size of 750 bytes.

Generally the following conclusions can be drawn from the study and simulations of Jim Zyren:

1.) The degree of interference experienced in any installation is dependent on local propagation conditions, the density of Bluetooth piconets, and Bluetooth piconet loading.

2.) IEEE 802.11b DSSS WLAN susceptibility to Bluetooth interference increases as a function of range from the DSSS wireless node to the DSSS AP

3.) IEEE 802.11b DSSS Hi Rate systems show graceful degradation in the presence of significant levels of Bluetooth interference

4.) Based on typical utilisation models of Bluetooth piconets, IEEE 802.11b High Speed WLANs show good reliability even in a fairly dense environment of Bluetooth piconets. (for details look at thepaper of Jim Zyren)

WLAN interferes Bluetooth

The impact of a 20dBm 802.11 Direct-Sequence WLAN system on a 0dBm Bluetooth link is studied in a paper written by Jaap C. Haartsen and Stefan Zürbes, Ericsson, with the title: „Bluetooth voice and data performance in 802.11 DS WLAN environment“ (Bluetooth-WLAN-interference.pdf). Thereby simulation results are presented.
For the interference studies, a typical office environment with 2 WLAN access points is assumed. WLAN terminals are uniformly distributed with a density of one terminal per 25 m2. A single access point serves 50 WLAN terminals. It is also assumed that a single Bluetooth piconet is associated with each WLAN terminal. Because of the distance between the WLAN terminals and the low Bluetooth transmit power the mutual interference between Bluetooth piconets is ignored.
The performance of the Bluetooth terminal is determined by the intended power received and the interfering power received, or the total C/I. This in turn will depend on

1. The distance between the Bluetooth receiver and Bluetooth transmitter

2. The distance between the Bluetooth receiver and the WLAN terminal transmitter
3. The distance between the Bluetooth receiver and the WLAN access point transmitter
When transmitting in its 22 MHz channel, the WLAN system effectively occupies about 17MHz of the 2.45GHz ISM band (20dBm bandwidth). The total amount of power transmitted amounts to 20dBm. When the Bluetooth receiver hops in the WLAN band, it filters out the Bluetooth hop bandwidth. For the Bluetooth receiver, the WLAN signal is regarded as white noise. Assuming a 0.85MHz noise bandwidth in the Bluetooth receiver, a filter suppression of 13dB is achieved. With a C/N required of 17dB @ 10-3BER, the required C/I towards a WLAN transmitter amounts to 4dB. The Bluetooth system transmits with a 0dBm power level. The 20dB transmit bandwidth is 1MHz.

For the range of interference, it is distinguished between voice and data performance. The Bluetooth data channel applies retransmission and can therefore cope with a higher packet erasure rate (PER) than voice. For the performance thresholds (the thresholds where still acceptable performance is experienced), PER=10% for data and PER=1% for voice was chosen. These two values must be considered with care since the user experience is largely determined by the time period the interference lasts. For example, a 2% PER for a period of 10 seconds in a voice connection will be more annoying to the user than a 10% PER in a period of 100ms.

Under normal traffic conditions in the WLAN (40 emails, 20 file transfers and 1000 Internet accesses per WLAN terminal in 8 hours, resulting in a total transfer of approximately 11 Mbyte per terminal), the Bluetooth voice user is not affected as long as his operating distance remains below 2m. If the operating distance increases to 10m, the probability that there is a noticeable interference on the link increases to 8%. The Bluetooth data link allows and experiences more degradation. A throughput reduction of more than 10% occurs with 24% probability at an operating distance of 10m. However, because of the limited frequency overlap of the WLAN and Bluetooth systems, the throughput reduction in the Bluetooth system can never exceed 22%, if only one WLAN system is installed.

Measurements

Measurements especially including mutual interferences with WLAN are shown in the measurement section "Radio Measurements".

Co-existance of Bluetooth and WLAN

The project members have performed tests which demonstrate that both Bluetooth and WLAN can co-exist in the same area, and even in the same devices (e.g. Laptop). Both technologies have their preferred area of usage, and these areas are almost complementary rather than competitive. WLAN dominates in the area of data connectivity, with the implementation of the wireless Ethernet. Bluetooth dominates in the "Personal Area Networking" domain, interconnecting all devices of your personal sphere, as e.g. Mobile Phone, PDA, Camera, Stereo, PC. Bluetooth has also implemented voice support (see own section on this topic).

The Specifications of both systems describe how the technology works, i.e. Bluetooth resp. WLAN protocol architecture.  While the WLAN architecture only covers the lower layers 1-3, Bluetooth covers the whole range from layer 1 (radio) to layer 7 (applications). A visualisation of the Bluetooth protocoll architecture is provided in figure 2, further information is available in the chapter on Bluetooth profiles.

Figure 2: The Bluetooth architecture, protocolls related to the OSI model

In summarising the above chapter, our assumption is a co-existance of WLAN and Bluetooth is the most likely scenario in the future. An assumption which is supported by the comparison of both technologies (see table 3)

Table 3: Comparison of Bluetooth and WLAN
 
Bluetooth
WLAN (802.11, 802.11b)
Data rate
4-700 kbps
4-6 Mbps
Range
10 m, up to 100 m
100 meter
# simultaneous users
7
10 – 50 depending on application
Frequency band
2,4 GHz
2,4 GHz
Transmitt power
1 mW, 2.5 or 100 mW 
100 mW
Interfaces
ADSL, Ethernet, ISDN, PSTN, USB, RS232
Primary Ehernet
Support for voice
Yes
Only VoIP
Type of clients
Inbuilt in PC, PDA, Mobile Phone, PC-card, CF-card, Memory Stick (Sony), SSD-card, USB dongle, RS-232 Dongle
Inbuilt in PC, PC-card, CF-card, RS-232 dongle and Ethernet dongle
Power management
Yes
Proprieatairy solutions
# of parallel systems
15 – 50 depending on application
3
Interoperability
Yes
Through WECA, not through standard

Comparison of different access technologies

A more detailed comparison of the different access technologies is available in the technical report