An 802.11 device and Bluetooth can interfere with each other when the 802.11 device operates on the 2.4 GHz band. All Bluetooth devices operate at the 2.4 GHz band. This section documents the technical details regarding the causes of interferences and solutions implemented in drivers, the 802.11 stack, and possible future enhancements.
Contents
802.11 spectrum use
802.11 defines channels on the 2.4 GHz band, each 20 MHz wide. Then, 802.11n allows for 40 MHz wide channels where you have a primary and an adjacent extension channel.
Bluetooth spectrum use
Bluetooth defines 79 channels for communication on the 2.4 GHz band each channel being separated by 1 MHz. The frequency range used is between 2.402 GHz and 2.480 GHz. Bluetooth's transmitted signal is spread across this 2.4 GHz band and the specification allows for 1600 frequency hops per second, the advantage being that since information is spread across a wide band of frequencies signals transmitted by other systems using a portion of the same frequency spectrum may appear as noise to only some of the frequencies used by Bluetooth device using frequency hopping and similarly only a portion of a Bluetooth transmission may interfere with signals transmitted by other systems.
802.11 and BT spectrum map
Center of frequency / freq-range (MHZ)
|
802.11 Channel number
|
BT Channel
|
2412 2402-2422
|
1
|
1-20
|
2417 2407-2427
|
2
|
5-25
|
2422 2412-2432
|
3
|
10-30
|
2427 2417-2437
|
4
|
15-35
|
2432 2422-2442
|
5
|
20-40
|
2437 2427-2447
|
6
|
25-45
|
2442 2432-2452
|
7
|
30-50
|
2447 2437-2457
|
8
|
35-55
|
2452 2442-2462
|
9
|
40-60
|
2457 2447-2467
|
10
|
45-65
|
2462 2452-2472
|
11
|
50-70
|
2467 2457-2477
|
12
|
55-75
|
2472 2462-2482
|
13
|
Not used
|
2484 2474-2492
|
14
|
Not used
|
Interference
Each 802.11 channel then equals to 20 Bluetooth channels. When communication is enabled on an Bluetooth device you will get interference when the Bluetooth device hops on to any of the 20 Bluetooth channels equivalent to your 802.11 channel. Even if a Bluetooth device hops at the max allowed frequency rate of 1600 frequency hops per second there are only 79 channels available so at this rate each channel will be used around 20 times in a second.
Coexistence techniques
To aid with 802.11 and Bluetooth interference issues several techniques have been implemented on Bluetooth device firmware and 802.11 devices. We elaborate on each of these techniques below.
Adaptive Frequency Hopping (AFH)
With this technique, required for Bluetooth devices following the 1.2 spec, Bluetooth devices will scan for fixed sources of interference and adapt its own frequency pattern to avoid those with interference. This works great in low noise environments in a noisy environment could end up disabling Bluetooth communication completely.
Channel skipping
When a Bluetooth device sits on the same system as an 802.11 device the 802.11 device can inform the Bluetooth device the 802.11 channel it is operating on and the Bluetooth device can preemptively disable communication on the respective Bluetooth channels. The advantage to this solution is the preemptive nature of preventing interference but you also end up limiting the Bluetooth's own spectrum.
Time Division Multiplexing
With this scheme an 802.11 device and Bluetooth device can take turns in using the spectrum for communication. The 802.11 device would use a WLAN_ACTIVE signal when it is active and a Bluetooth device can use a BT_PRIORITY to categorize the priority of its different own signals.
Typically signals between an 802.11 and Bluetooth device are triggered using GPIO lines.
2-wire Bluetooth Coexistence
Under the 2-wire Bluetooth Coexistence scheme two signals are supported:
- WLAN_ACTIVE - signals when an 802.11 device is active, 802.11–>BT
- BT_PRIORITY - categorizes the Bluetooth signal, BT–>802.11
3-wire Bluetooth Coexistence
The 3-wire Bluetooth Coexistence scheme extends 2-wire scheme with a new signal from the bluetooth device:
- BT_STATE: signals the presence of a Bluetooth transmission.
Interoperability
Apart from AFS and channel skipping techniques Bluetooth coexistence is typically tested with bundled 802.11 and Bluetooth devices. This becomes more evident with 2-wire and 3-wire which relies on GPIO pins for signaling.
Driver Bluetooth coexistence
This section refers to Bluetooth coexistence support on existing 802.11 Linux drivers.
Ideas for development
Once the Frequency broker is implemented add a trigger on an 802.11 channel switch to the disable the respective Bluetooth channels. This would be a general cheap BT-coex scheme independent of device interoperability.
References
-
Wi-Fi and Bluetooth: Enabling CoexistenceJim Lansford, Ron Nevo, and Brett Monello
The coexistence of Bluetooth and Wi-Fi (IEEE 802.11b) requires particular attention to simultaneous operation of both systems in very close proximity.
The 2.4 GHz Industrial, Scientific, and Medical (ISM) band is poised for strong growth. Fueling this growth are two emerging wireless technologies: wireless personal-area networking (WPAN) and wireless local-area networking (WLAN). The dominant WPAN technology is a short-range wireless technology called Bluetooth. Cahners In-Stat (Scottsdale, AZ) predicts the market for Bluetooth devices will reach 800 million units by 2004.1
Designed principally for cable-replacement applications, most Bluetooth implementations support a range of up to ~10 m and speeds of up to 700 Kb/sec for data or isochronous voice transmission. Bluetooth is ideal for applications such as wireless headsets, wireless synchronization of personal digital assistants (PDAs) with computers, and wireless peripherals such as printers or keyboards.
WLAN has several technologies competing for dominance; however, based on current market momentum, it appears that Wi-Fi (IEEE 802.11b) will prevail. Wi-Fi offers speeds of 11 Mb/sec and covers a range of up to 100 m. With WLANs, applications such as Internet access, e-mail, and file sharing can now be implemented with new levels of freedom and flexibility. The market for WLAN shipments is predicted to reach more than 5 million units in 2004, implying an installed base of nearly 20 million systems.2
WPAN and WLAN are complementary rather than competing technologies. Moreover, with both expecting rapid growth, collocation of Bluetooth and Wi-Fi devices will become increasingly likely. Because both technologies occupy the 2.4 GHz frequency band, there is potential for interference between the two technologies. Coexistence of the two technologies has become a key topic for analysis and discussion throughout the industry. In this article, the term coexistence means that the wireless systems can be collocated without significantly affecting the performance of any device.3
Previous Work
Analysis of interference between Wi-Fi and Bluetooth devices is not new, as indicated by the literature such as Kamerman.4 Early attempts to quantify the mutual interference effects have been based on simple geometric models of Bluetooth deployment rather than actual usage models.5,6 In Ennis, the investigation focused on the problem of calculating the probability of an overlap, in both time and frequency, of a continuous sequence of Bluetooth packets and an IEEE 802.11b direct sequence 11-Mb/sec packet.7Relative power levels between the Bluetooth and IEEE 802.11b packets were not considered. Zyren and others made several refinements on previous assumptions.8–10 These efforts, however, did not examine in detail the ramifications of the physical (PHY) layer such as hopping, spectral masks, and filter selectivity, nor did they discuss implementation. In addition, the geometries studied did not necessarily correspond to practical usage models.
Results presented by Golmie modeled both PHY and medium access controller (MAC) behaviors. Such modeling is necessary to predict performance accurately. This article uses some of the same approaches to modeling, providing a firm foundation from which to evaluate the effectiveness of interference-reduction techniques.
Interference Basics
With characteristics such as low cost, modest speed, and short range (<10 a="a" as="as" bluetooth="bluetooth" cable-replacement="cable-replacement" can="can" designed="designed" i="i" m="m" nbsp="nbsp" radio-frequency="radio-frequency" support="support" technology.="technology." was="was">piconets
Sharing the Same Frequency Band
Wireless communication systems use one or more carrier frequencies (frequency bands) to communicate. Bluetooth and Wi-Fi share the same 2.4 GHz band, which under Federal Communications Commission (FCC) regulations, extends from 2.4 to 2.4835 GHz. Under the ISM band rules defined in FCC Part 15.247, this frequency band is free of tariffs.11 It is license exempt in Europe. However, systems must operate under certain constraints that are supposed to enable multiple systems to coexist in time and place.
A system can use one of two methods to transmit in this band; both are spread-spectrum techniques. Frequency-hopping spread spectrum (FHSS) enables a device to transmit high energy in a relatively narrow band, but for a limited time. Direct-sequence spread spectrum (DSSS) allows a device to occupy a wider bandwidth with relatively low energy in a given segment of the band, and it does not hop.
As discussed earlier, Bluetooth selected FHSS, using 1-MHz-wide channels and a hop rate of 1600 hops/sec (625 microseconds in every frequency channel). Bluetooth uses 79 different channels in the United States and most of the rest of the world. IEEE 802.11b (Wi-Fi) opted for DSSS, using 22 MHz of bandwidth (passband) to transmit data with speeds of up to 11 Mb/sec. A Wi-Fi system can use any of 11 22-MHz-wide subchannels across the allocated 83.5 MHz of the 2.4 GHz frequency band. A maximum of three Wi-Fi networks can coexist without interfering with one another. Geographies outside of the United States may support more or fewer than 11 selectable subchannels. However, regardless of the portion of the band in which Wi-Fi operates, sharing with Bluetooth is inevitable. Two wireless systems using the same frequency band would have a high propensity to interfere with each other.
Bluetooth and Wi-Fi Interference Cases
If Bluetooth and Wi-Fi operate at the same time in the same place, they will interfere (collide) with each other. Specifically, these systems transmit on overlapping frequencies, creating in-band colored noise for one another. The sidebands of each transmission must also be accounted for. Interference between Bluetooth and Wi-Fi occurs when either of the following is true:
- An Wi-Fi receiver senses a Bluetooth signal at the same time a Wi-Fi signal is being sent to it. The effect is most pronounced when the Bluetooth signal is within the 22-MHz-wide passband of the Wi-Fi receiver.
- A Bluetooth receiver senses a Wi-Fi signal at the same time a Bluetooth signal is being sent to it; the effect is most pronounced when the Wi-Fi signal is within the passband of the Bluetooth receiver.
Interference Measurements
It is important to understand a few simple measurements that demonstrate the potential magnitude of the interference issue in scenarios that involve simultaneous operation.
 |
Figure 1. Geometry of measurement environment.
|
Measurement Environment. A geometric arrangement of Bluetooth and Wi-Fi devices (shown in Figure 1) was tested. The configuration was intended to be representative of a laptop (a device which needs simultaneous operation and collocation) equipped with collocated Wi-Fi and Bluetooth interacting simultaneously with a Wi-Fi access point and another Bluetooth node. Ganymede's Chariot, a LAN traffic excitation program used by the Wireless Ethernet Compatibility Alliance (WECA) for certification of IEEE 802.11b interoperability, was used to measure Wi-Fi throughput.
In this test, Chariot drove IEEE 802.11b RoamAbout PC cards (Cabletron Systems) with Ethernet packet sizes of 1500 bytes (typical for such data-transfer applications). Data transfer was from LT1 to LT2, and RF output power was 30 mW. Measurements were captured at varying distances from LT1 to LT2. The distance between the collocated Wi-Fi and Bluetooth (LT2 to BT1) was fixed at 10 cm, and the second Bluetooth node (BT2) was located 1 m from LT2. The 10-cm spacing is less than the maximum achievable inside a laptop, but greater than what could be achieved in a single PC card. The two Bluetooth nodes were laptops with Digianswer PC cards that ran DH5 data transfers from BT1 to BT2 at an RF power output of 1 mW.
Measurement Results. Figure 2 presents two sets of measurement results that capture Wi-Fi throughput as a function of received signal strength at the Wi-Fi station (LT2). The diamond line shows Wi-Fi performance with all Bluetooth devices turned off (baseline). The square line shows Wi-Fi throughput with Bluetooth active.
 |
Figure 2. Measured throughput of Wi-Fi in the presence of Bluetooth (IEEE 802.11b Mb/sec versus received signal).
|
Ultimately, it is important to understand the effect on Wi-Fi performance over distance. However, given that the effect of distance is highly dependent upon the actual physical environment, received signal strength indicator (RSSI) was selected. In this way, results could be extended across environments by making simpler physical energy-level measurements rather than having to reproduce and execute full test suites. Table I shows the received signal strengths and distances for a particular office environment with cubicles. Using RSSI enables interpretation of data without having to account for the effects of selective fading.
In the scenarios measured, even Wi-Fi stations with <5 access="access" free="free" from="from" m="m" of="of" point="point" space="space" suffer="suffer" their="their">25% degradation in throughput. This degradation exceeds 50% by the 30-m mark. In an office environment with cubicles, the range associated with each throughput level would be reduced significantly. When cubicles must be penetrated, Wi-Fi loses nearly one-third of its expected throughput within the first couple of meters. Erosion of performance exceeds 50% with stations <8 access="access" although="although" as="as" bluetooth="bluetooth" can="can" coexistence="coexistence" degrade="degrade" deleterious="deleterious" demonstrate="demonstrate" effect="effect" experiment="experiment" focus="focus" from="from" however="however" is="is" issues="issues" m="m" not="not" of="of" on="on" p="p" particular="particular" performance.="performance." point.="point." s="s" simulations="simulations" subsequent="subsequent" substantially="substantially" that="that" the="the" their="their" this="this" voice="voice" was="was" wi-fi.="wi-fi." wi-fi="wi-fi">
| Received SignalStrength (dBm) | Degradation vs.Baseline (%) | Distance (m)Free Space | Distance (m) andNumber of Cubicle Walls |
| –25.62 | 14.33 | 1 | — |
| –31.64 | 16.26 | 2 | — |
| –35.17 | 19.23 | 3 | — |
| –39.61 | 24.01 | 5 | — |
| –42.53 | 25.11 | 7 | — |
| –46.93 | 30.11 | 10 | — |
| –52.74 | 34.29 | 15 | 2 m; 1 wall |
| –56.86 | 32.60 | 20 | 4 m; 2 walls |
| –62.67 | 62.88 | 30 | 6 m; 3 walls |
| –69.99 | 53.78 | 50 | 8 m; 4 walls |
| –74.82 | 63.63 | 70 | — |
| –79.93 | 95.50 | 100 | 20 m (down a corridor); 2 office walls |
Table I. Received signal strength versus distance.
It should be stressed that interference between radio systems is highly variable and depends on a number of factors, primarily the geometry of the nodes. Given the nature of radio-wave propagation and the practicalities of receiver design, it is always possible to construct scenarios that will give pathologically poor performance (or unrealistically excellent performance). The scenario outlined in the preceding discussion, as well as the simulation scenarios that follow, represent neither extreme. They should, however, compel WPAN and WLAN developers to work toward solutions.
Interference Simulation
Using knowledge of the characteristics of Bluetooth and Wi-Fi, a detailed simulation environment was created to make quantitative assessments of the mutual impact of interference. This simulation was designed to reinforce the understanding of interference mechanisms and to help develop and evaluate high-impact coexistence techniques.
Simulation Overview. Mobilian Corp. created the simulation for the purpose of characterizing Bluetooth and Wi-Fi interference effects and for identifying solutions that would enable coexistence and simultaneous operation. This highly flexible C program accurately models the behavior of both the PHY and MAC of both Bluetooth and Wi-Fi. In fact, most of the key parameters discussed can be varied to simulate different scenarios.
The typical scenario involves a Sim-OP laptop interacting with a Wi-Fi access point and a Bluetooth node. Bluetooth piconets and Wi-Fi stations can be added. Distances between access point and laptop, between laptop and Bluetooth node, and between laptop and other Bluetooth piconets can be varied easily. Wi-Fi data rates and packet sizes, along with Bluetooth operation modes, can also be defined and varied.
Scenario Setup and Calibration. To maximize the credibility of the simulation results provided, the simulation was calibrated to the measurement data shown in Figure 2. Once the proper parameters are set in the simulation, accurate projections of the behavior of the modeled systems can be produced and varied over additional scenarios. As with the measured data, in each simulated scenario, the Wi-Fi access point– to–Wi-Fi station distance varies. The objective of the first scenario was simply to replicate the results obtained through the actual experiment as a way of validating the simulation setup. The distance from the Wi-Fi station (STA) to the collocated Bluetooth node is fixed at 10 cm. The complementary Bluetooth node is located 1 m away. Table II documents additional parameter values used to produce the simulation results.
 |
Figure 3. Calibration of simulation and experimental trials.
|
Note that this model does not directly account for multipath, although the path-loss term beyond 8 m does take into account some of the range reduction due to reflections. As with any empirical model, it represents mean path-loss values and cannot predict the specific frequency-selective effects that have a significant effect on frequency-hopping systems. Such detailed simulations are highly site specific and must employ models that predict exact signal paths, which cannot be generalized and typically require several hours to run. An empirical model such as the one used here is a highly useful tool for predicting performance that can be tied to average measurements.
Simulation Scenarios. After establishing a calibrated baseline, additional parameters were varied to further understand the effect of various key parameters. In particular, simulations were conducted that increased the intensity of Bluetooth activity (from DH5 to SCO HV1), added more Bluetooth piconets, and increased Wi-Fi transmission power from 30 to 100 mW.
 |
Figure 4. The impact of Bluetooth DH5 versus HV1.
|
 |
Figure 5. Basic geometry of the office scenario.
|
 |
Figure 6. An office scenario with varying Wi-Fi power levels.
|
Effect of Wi-Fi on Bluetooth
It is also important to examine the degradation of Bluetooth performance in the Wi-Fi environment. Figure 7 shows the throughput reduction of the Bluetooth node as a function of the STA-Bluetooth pair's distance from the Wi-Fi access point.
 |
Figure 7. Throughput reduction of Bluetooth (synchronous-connection-oriented links) in the presence of Wi-Fi (IEEE 802.11b).
|
Because the Wi-Fi system looks like a broadband jammer to the Bluetooth node, one would expect to see voice-packet failures occur more or less randomly until the access point is sufficiently far away (that is, the signal-to-noise ratio increases) to let the capture effect block the interference. Because the Bluetooth receiver is nonlinear, this transition should occur sharply, and the simulation indicates that it does. It also is interesting to note that the relatively short (<150 -microsecond="-microsecond" access="access" acknowledgements="acknowledgements" bluetooth="bluetooth" cause="cause" do="do" ethernet="ethernet" from="from" however="however" in="in" longer="longer" loss.="loss." much="much" not="not" p="p" packet="packet" packets="packets" point="point" s="s" seem="seem" severe="severe" significant="significant" station="station" the="the" to="to">
Parameter
|
Wi-Fi Station (STA) and Access Point (AP)
|
Bluetooth Nodes
|
| Path loss model | Lp = –40.0 – 20log(d), when d ¾ 8 m Lp = –58.5 – 33log(d/8), when d > 8 m |
Lp = –40.0 – 20log(d), when d ¾ 8 m
Lp = –58.5 – 33log(d/8), when d > 8 m |
| Environmental description | Open air |
Open air
|
| Transmit power | 30 mW |
1 mW
|
| Transmit mask | 11 < |f – fc|< 22 : –30 dB 22 < |f – fc| : –50 dB |
0.5 < |f – fc| ¾ 1.5 : –20 dB
1.5 < |f – fc| ¾ 2.5 : –40 dB 2.5 < |f – fc| ¾ 3.5 : –60 dB 3.5 < |f – fc| : –infinity |
| Compression point | Ignored |
Ignored
|
| Receiver sensitivity | –83 dBm |
–91 dBm
|
| Receiver bandwidth (passband) | 22 MHz |
1 MHz
|
| Receiver attenuation; Adjacent/alternative channel selectivity | 11 <|f – fc| < 12: –12 dB 12 <|f – fc| < 22: –36 dB 22 <|f – fc| : –56 dB |
0.5 < |f – fc| ¾ 1.5 : –11 dB
1.5 < |f – fc| ¾ 2.5 : –41 dB 2.5 < |f – fc| : –51 dB |
| Mode/Data rate/Packet length | Asymmetric packet flowfrom AP – STA. STA sends Layer 3 ACKs only. 11-Mb/sec, 1500-byte packets |
Varies by scenario
DH5 or HV1 |
| MAC behavior | Reduction in data rate and fragmentation mechanisms disabled |
—
|
Table II. Simulation setup parameters.
Interference-Reduction Techniques
As the physical measurements and simulation results demonstrate, the coexistence problem is significant. A number of techniques can be employed to reduce the interference between Wi-Fi and Bluetooth. Several are under investigation by the IEEE 802.15.2 working group. These techniques can be grouped into four general categories:
- Regulatory and standards (spectrum-usage regulations and specifications in standards bodies).
- Usage and practices.
- Technical approaches (general system approaches, driver layers, MAC layers, and physical layers).
- Alternate frequency bands (5+ GHz).
Regulatory Proposals. Under current (April 2001) FCC rules, Bluetooth is required to hop over almost the entire ISM band from 2.400 to 2.4835 GHz. Further, it must occupy a minimum of 75 MHz, and each channel in Bluetooth is 1-MHz wide. The Bluetooth specification actually calls for 79 channels, so it must cover virtually the entire band as it pseudorandomly hops around.12 This band coverage is the reason for the inevitable time-frequency collisions between Bluetooth and Wi-Fi.
A rule change that would allow wireless systems using 1-MHz-wide channels to hop over only part of the ISM band has been proposed; for example, a piconet might hop over only a segment of the band. Although a host of technical details remain to be worked out, this proposal would in principle allow Wi-Fi and Bluetooth to completely avoid each other in some scenarios. However, this solution may not help much in dense environments such as the fully loaded enterprise scenario where two or three Wi-Fi networks on different frequency bands are coexisting.
Because rule changes can often take a significant amount of time and details of the final Report and Order are impossible to predict in advance, it is difficult to assess the timing and effectiveness of these proposals. Based on experience in the cordless telephone industry, there is reason to believe that a properly designed adaptive hopping system can improve coexistence in many cases. However, given filter selectivity, it is not known whether adaptive hopping will allow simultaneous operation when collocated.
Standards and Industry Bodies. As awareness of coexistence issues has grown, progressive groups within the industry have begun to address the problem and look for solutions. The most active group is the IEEE 802.15.2 working group, which issued a formal call for proposals in September 2000. This group is expected to publish a recommended practices document in 2001 that will address a range of coexistence solutions. Recommended practices will be published for collaborative and noncollaborative coexistence; the former means that there is direct communication between the two systems, whereas the latter means they must indirectly infer the interference environment. The Bluetooth special interest group has a Wireless LAN Coexistence Working Group that is also actively involved in seeking solutions to the coexistence problem. WECA also has an ad hoc task force addressing coexistence issues. It plans to publish white papers that will outline the magnitude of the problem and recommended solutions.
Usage and Practices
One solution some companies have taken toward Bluetooth–Wi-Fi coexistence is to ban Bluetooth. One could also imagine a ban on Wi-Fi in environments where Bluetooth is considered mission critical. This is not an especially workable solution, because people will want to use the tools that best fit their needs.
At the level of the individual user, modal operation of Bluetooth and Wi-Fi could be practiced. This approach is reasonable where Bluetooth applications are exercised only sporadically and for short durations (for example, a daily sync between a PDA and a desktop). However, as densities of Wi-Fi and Bluetooth both grow, interference to and from neighboring cubicles becomes a relevant consideration and tends to undermine such modal-usage models. More fundamentally, any proposal that requires such a high degree of user awareness and behavior modification is unlikely to succeed.
Technical Approaches
General System Approaches. Because of the role of signal-to-noise energy levels in determining packet loss, it is tempting to explore the role of transmission power in improving coexistence. Mobilian simulations do demonstrate that lowering the power levels in the collocated Bluetooth node does not change the basic shape of the Wi-Fi performance degradation curve, but rather shifts this curve to the right, increasing the range over which any given throughput level is possible. Many Bluetooth usage models (involving a notebook computer, for example) require relatively short-range interaction with other Bluetooth devices such as PDAs and peripherals. Given these limited range requirements, Bluetooth systems with lower or variable power could be viable and would lessen the interference impact on Wi-Fi. Likewise, power control techniques in Wi-Fi are being investigated by several companies; this effort will be necessary in the MAC for 5-GHz WLANs sold in Europe, where the European Telecommunication Standards Institute (ETSI) requires transmit power control (TPC).
Driver Layer. Some discussion within the industry has centered on using the software layers above the MAC to switch between the Bluetooth and Wi-Fi systems in devices that have both installed. This approach is attractive for those systems that need to communicate only with either, but not both, Wi-Fi or Bluetooth in a time-critical application. This approach would support limited transfers in a ping-pong fashion, but would not be able to support Wi-Fi traffic while Bluetooth voice was active in the piconet. Bluetooth peripherals such as human-input devices also would not function well while the WLAN was active. Overall, this solution is limited in its usefulness.
MAC Layer. The MAC layer is an attractive place to focus attention on improving the coexistence between Bluetooth and Wi-Fi, because that is where such techniques as listen-before-talk functions are implemented. Bluetooth's very fast and unpredictable hop pattern makes it difficult to base the operation of Wi-Fi on either listen-before-talk techniques or on a history of previous failures.
In addition, there is no mechanism for Bluetooth and Wi-Fi to exchange information directly about future activities. Neither has the ability to effectively plan around the other. If rule changes such as those discussed in the section on regulatory approaches were approved, some method for minimal mutual identification and information exchange would likely be required to support implementation.
The MAC layer is where data rates are determined, so this is the place to resolve data-rate versus packet-size tradeoffs. Because the MAC layer comprises digital hardware and software, techniques employed there tend to be relatively inexpensive to implement. However, not all problems can be solved in the MAC. For example, the MAC has no control over timing under some conditions, such as when a Wi-Fi node is required to respond with an ACK within a few microseconds of the successful reception of a packet. However, given potentially deleterious MAC behavior in the face of Bluetooth interference, more-intelligent back-off and fragmentation algorithms might improve Wi-Fi throughput.
Physical Layer. Collisions actually happen at the PHY layer, and, as discussed earlier, a number of parameters can be influenced via design decisions. Some time-frequency collisions cannot be avoided unless PHY-layer techniques are used. For example, the IEEE 802.11b specification requires that an ACK be transmitted within a few microseconds after a packet is successfully received. If the same station is also transmitting a Bluetooth packet at that time, then the node that is expecting the ACK may be jammed by the Bluetooth signal. Only by the use of signal-processing techniques in the PHY layer can the Bluetooth signal be excised from the Wi-Fi passband so that the ACK can be successfully processed. PHY-layer techniques tend to directly affect system costs more than MAC-layer techniques, so this must be considered.
Alternate Frequency Bands
Some in the industry have positioned coexistence problems in the 2.4 GHz frequency band as motivation to hasten the migration to 5-GHz WLAN standards such as IEEE 802.11a and HiperLan2. The issues of WLAN at 5 GHz certainly warrant a separate article, but the relevant points are summarized here:
The physics of radio-wave propagation introduce a path-loss penalty of 6.9 dB going from 2.4 to 5.3 GHz. This means that almost five times as much RF power is required to cover the same distance—all other factors being equal. Because RF power amplifiers at 5 GHz are relatively expensive, IEEE 802.11a systems will either cover a smaller distance, cost more, or both.
RF propagation through barriers such as walls is also somewhat poorer at 5 GHz, so link budgets may suffer additional losses.13 This loss suggests that an additional power penalty above and beyond the path-loss effect described earlier is likely.
Strictly speaking, HiperLan2 and its predecessor, HiperLan1, are the only systems that are currently legal in Europe.14 Although HiperLan2 and IEEE 802.11a have very similar physical (radio) layers, they are very different at the MAC layer, so another standards battle is possible. Under current ETSI rules, IEEE 802.11a is not legal for use in the 5 GHz band in Europe. Efforts are under way in the IEEE 802.11h Task Group to add features to the IEEE 802.11a/b MAC specification that will allow these WLAN systems to meet the ETSI requirements for TPC, possibly allowing IEEE 802.11a to become legal for use in Europe.
FCC regulations do not prohibit other kinds of systems in these bands; both PAN systems and microwave ovens are being developed for portions of this band. This could lead to coexistence issues in this band as well, in addition to the IEEE 802.11a and HiperLan2 issues.
Both HiperLan2 and IEEE 802.11a use a spectrally efficient signal-modulation technique called orthogonal frequency division multiplexing (OFDM), which is also multipath resistant. Unfortunately, OFDM also has a very high ratio of its peak-to-average value compared to phase-shift keying (PSK), which is used in IEEE 802.11b. The result of this high ratio is that the RF power amplifier (one of the most expensive components in the system) must be designed to accommodate this large signal swing, which drives up cost and power consumption, unless RF power is scaled back (shorter range).
It is very difficult to manufacture 5-GHz systems on conventional printed circuit board material (FR4) because the loss tangent and dielectric constants yield poor performance. Consequently, more-exotic (and more-expensive) materials are generally needed.
These issues notwithstanding, wireless LAN deployment in the 5 GHz band offers the potential to dramatically increase capacity in a WLAN deployment when capacity is expressed as megabits per second per cubic meter.
Conclusion
Coexistence, and ultimately simultaneous operation, between Wi-Fi and Bluetooth technologies is a desirable goal. Both technologies are expected to grow rapidly over the next few years, offering new levels of portability and convenience, and many critical usage models require collocation and simultaneous operation of both standards in the same device. Systems-level approaches that address coexistence through the use of antenna, PHY, and MAC techniques offer the potential to dramatically reduce, if not eliminate, interference between these two systems. Such robust wireless-system design technology will become increasingly important in the unlicensed bands as Bluetooth, Wi-Fi, and other unlicensed wireless technologies proliferate.
A joint proposal by Mobilian Corp. and Symbol Technologies was approved as recommended practice by the IEEE 802.15.2 Task Group in March 2001 for simultaneous operation when collocated. Systems that implement this specification are expected to appear in the market in 2001.
References
1. Bluetooth 2000: To Enable the Star Trek Generation, Report #MM00-098W (Scottsdale, AZ: Cahners In-Stat, July 2000).
2. Enterprise Wireless LAN Market Analysis, Report #LN00-11WL (Scottsdale, AZ: Cahners In-Stat, June 2000).
3. D Cypher, "Coexistence, Interoperability and Other Terms," IEEE 802.15: 99/134r2P802-15TG2, presentation to IEEE P802.15 Working Group for Wireless Personal Area Networks, September 2000.
4. A Kamerman, "Coexistence between Bluetooth and IEEE 802.11 CCK Solutions to Avoid Mutual Interference," presentation to Lucent Technologies Bell Laboratories, Murray Hill, NJ, January 1999. Also available as IEEE 802.11-00/162, July 2000.
5. N Golmie and F Mouveraux, "WPAN Coexistence Performance Evaluation: MAC Simulation Environment and Preliminary Results," IEEE 802.15-00/066r0, March 2000.
6. S Shellhammer, "Packet Error Rate of an IEEE 802.11 WLAN in the Presence of Bluetooth," IEEE 802.15-00/133r0, May 2000.
7. G Ennis, "Impact of Bluetooth on 802.11 Direct Sequence," IEEE 802.11-98/319, September 1998.
8. J Zyren, "Extension of Bluetooth and 802.11 Direct Sequence Model," IEEE 802.11-98/378, November 1998.
9. J Zyren, "Reliability of IEEE 802.11 Hi Rate DSSS WLANs in a High Density Bluetooth Environment," in Wi-Fi.org [on-line], 8 June 1999; available from Internet: http://www.wi-fi.org/ downloads/Coexistence_Paper_Intersil_Aug18.pdf.
10. J Zyren, "Reliability of IEEE 802.11 WLANs in Presence of Bluetooth Radios," IEEE 802.15-99/073r0, September 1999.
11. Code of Federal Regulations, 47 CFR 15.247. Volume 1, Part 15, Subpart C (Intentional Radiators), Section 247. US Government Printing Office. In the October 1, 1998, revision of 47 CFR 15.247, the relevant pages are 690–692. Available from Internet: http://www.access.gpo.gov/nara/waisidx_99/47cfr15_99.html.
12. Specification of the Bluetooth System, version 1.0B, December 1, 1999.
13. G Durgin, TS Rappaport, H Xu, "Measurements and Models for Radio Path Loss and Penetration Loss in and Around Homes and Trees at 5.85 GHz," IEEE Transactions on Communications 46, no. 11 (1998): 1484–1496.
14. ETR 0230002 V0.2.0 (1999-04), "Broadband Radio Access Networks (BRAN); High Performance Radio Local Area Networks (HIPERLAN) Type 2; System Overview," European Telecommunications Standards Institute, Brussels.
Jim Lansford, PhD, is vice president of business development for Mobilian Corp. (Hillsboro, OR). Ron Nevo is an engineering director for Mobilian, and Brett Monello is a vice president for Mobilian and country manager for the company's Israeli operations. They can be reached at 503/681-8600 or via http://www. mobilian.com.
Illustration by Hermann/Starke (Corbis)
Thanks to :
No comments:
Post a Comment