Bluetooth and WiFi integration: Solving co-existence challenges

musang_pulut

The two most popular wireless technologies are Bluetooth

musang_pulut

Interference mitigation
Designing consumer electronics that include WLAN and Bluetooth can be quite problematic, particularly when a simultaneous mode of operation is required. Bluetooth and 802.11b/g WLAN systems operate in the 2.4 GHz band, and both technologies use a significant portion of the available spectrum. (Figure 2).
Interference occurs when a Bluetooth and a WLAN device are in close proximity and attempt to transmit and receive wireless signals at the same time. The two technologies use different methods for signal transmission: carrier sense multiple access (CSMA) and frequency hopping spread spectrum. The former is used by an 802.11b/g transceiver to listen for a clear channel before transmitting a signal that is approximately 20 MHz wide and typically occupies one of up to three possible non-overlapping channels spaced 25 MHz apart.
In contrast, Bluetooth uses frequency hopping spread spectrum to hop between 79 1 MHz-wide channels at a rate of 1600 hops per second, delivering short time division multiplexed packets with each hop. A Bluetooth connection is made when one device initiates a connection and becomes the master of the piconet. If the address is known, the device transmits a page message. If the address is not known, an inquiry message is sent, followed by a page message. Once synchronized, two Bluetooth-enabled devices connect, with each assuming a unique media access control (MAC) address. A simple calculation demonstrates that a Bluetooth transmitter will output a signal that collides with an 802.11b/g signal approximately 25% of the time.
The effect of this co-channel interference depends on the relative strengths of both the signals and the transmit length and duty cycles of the data packets. Various analytical and simulation studies have investigated interference scenarios whereby one or both types of signals are adversely affected2,3. The specifications have a degree of interference resilience resulting from their signal design: direct sequence spread spectrum (DSSS) for 802.11b, orthogonal division frequency multiplexing (OFDM) for 802.11g, and frequency hopping for Bluetooth. They also use protocols based upon data packet retransmission and data rate reduction. However, these measures can result in a dramatic reduction in data throughput, which can negatively impact performance in some applications. Take, for example, Bluetooth audio transmission or VoIP over WLAN, where packet error rates greater than a few percent would cause intolerable audio delays or dropped calls.
The Bluetooth specification version 1.2 includes adaptive frequency hopping (AFH) as a way of allowing Bluetooth devices to detect and avoid interference. Using AFH, a channel can be classified as good or bad so that bad channels are avoided and replaced in the hopping sequence by pseudo-randomly selecting out of the remaining good channels. Once a Bluetooth device determines that a WLAN device is operating within the 2.4 GHz band, the frequency hopping channels that overlap are designated as bad and avoided. The use of AFH in the United States was made possible by the Federal Communications Commission (FCC) in 2002 when it began allowing frequency hopping between a minimum of 15 and maximum of 75 channels. In other regulatory domains such as the European Telecommunications Standards Institute (ETSI), frequency hopping is also allowed with a minimum of 20 channels. This is the minimum number chosen for AFH

musang_pulut

Co-location interference
Unfortunately, techniques such as AFH are designed specifically to detect and avoid interference for 2.4 GHz devices rather than to enable co-existence. AFH as a stand-alone technique is insufficient when Bluetooth and 802.11 devices are co-located in the same design (Figures 3 and 4). This is due in large part to the fact that WLAN devices must have higher output power to support reliable high data rate transmission of Internet, voice data and video over longer distances.
In products where WLAN is co-located with Bluetooth Class 2 or Class 3 capability, the WLAN transmitter can have as much as a +20 dB higher output power. This increases the bandwidth of appreciable interference power relative to a non-co-located scenario, thereby minimizing the number of frequency-hopping channels available, and limiting the effectiveness of spectral spreading. To put this in perspective, the sole use of AFH in a dual-mode portable terminal will not support a voice conversation using a Bluetooth headset while the 802.11b/g device is trying to upload a packet.
Another consideration is the emerging use of Class 1 Bluetooth, which also uses a power amplifier for wireless transmission over a longer 100-meter range. This specification requires +20 dBm output power, which when co-located in a dual-mode terminal, can disrupt WLAN throughput performance. In fact, even with +40 dB isolation between radios, interference from a Bluetooth Class 3 transmitter, which has 0 dBm output power can degrade the throughput of an 802.11b device.
Apart from the overlapping transmitter spectrum, receiver desensitization is a serious concern. Due to the close proximity of the two transceivers, signals being transmitted from one RF chain can be large enough to saturate the receiver front-end of the other RF chain. This can occur regardless of the frequency offset, since all channel-specific filtering occurs at a lower IF, or at the baseband for the most common direct conversion receiver architectures. If transmissions between the two systems are not coordinated at a high level, then the saturation could occur anywhere in time during the reception of a data packet. During reception of a packet the receiver automatic gain control is set and the wanted signal could easily be 70 dB or 80 dB below the unwanted transmitter signal. Even if the transmitter signal itself is not a strong blocker, its wideband noise floor may be high enough to dominate the receiver noise floor.
The phase noise power of the receiver local oscillator may be strong enough at the transmitter frequency offset such that a co-channel interferer is formed on the downconverted signal by a process known as reciprocal mixing. Non-linearities in both co-located transmitters operating simultaneously can cause cross-modulation components at certain forbidden frequencies and at high enough power spectral densities to violate FCC and ETSI limits on wideband spurious generation. Similarly, transmitter harmonics can interfere with higher frequency restricted bands and 5 GHz 802.11a devices.
Isolation between co-located radios is the most effective way to reduce the level of transmitter blocker signals. This is usually limited in practice by small form factors and finite isolation effects afforded by antenna orientation and layout. Additional filtering on each transmitter is the most effective way of reducing interference levels, but this also impacts the bill of materials size and cost. The other disadvantages of extra filtering are that its losses effectively reduce transmitter efficiency and linearity at a given output power. Cross modulation components can be reduced by increasing the transmitter linearity but at the cost of efficiency, which again is undesirable for portable devices.
In many cases, the Bluetooth and WLAN functions are served with an architecture that includes physically separated baseband, RFIC, front-end and antenna blocks. The radios are on separate sides of a multilayer PCB, separated by ground plane and shielded

musang_pulut

Coordination measures for co-existence
Various proprietary mechanisms exist to coordinate radio activity to prevent simultaneous operation of the co-located devices 4,5. The approaches vary in detail but essentially act to interleave operation in order to make the operation appear simultaneous. The techniques address the scheduling and priority setting of the two systems, making trade offs on transmission duty cycle, idle times, and packet type (data/beacon/paging). The packets of one system can be sent while the other is idle and vice versa; the end effect is to deliver reliable communication on both systems with a negligible loss of throughput. Three examples of coordination approaches are highlighted below.
Dual-mode radio switching is the easiest co-existence mechanism to implement. This method works by completely suspending the operation of one radio while the other is operational. There are two ways to implement this. The first approach requires turning off the non-operating radio without signalling to other nodes in the network. The drawback of this approach is that it can reduce performance in some cases below that which is achieved with no coordination. The second method signals other network nodes that the radio is suspended. The reported performance is better than shutting off the radios, but is still about 60% lower than that of unhindered radios. Neither method supports switching while Bluetooth voice links are in operation.
Driver-level transmit switching generally describes an approach in which application transmit requests are mediated at the driver level, thereby avoiding simultaneous transmission. This approach degrades throughput because only one radio is active at a time, and it causes problems with packet collisions. As a result, systems using driver-level transmit switching can suffer from dropped packets, interference and potential user difficulties when simultaneously transmitting one protocol and receiving another. As with dual-mode radio switching, this approach does not switch quickly enough to support Bluetooth SCO links, and will also have difficulties mitigating the interference from Bluetooth piconet master/slave polling activities.
The most effective means to providing performance levels approaching no-interference scenarios is MAC-level switching. MAC-level switching is a collaborative technique, which exchanges information between the two protocols at the MAC level, and manages transmit and receive operations. Since MAC-level switching is performed in the baseband, it is able to switch between protocols at a much faster rate than driver-level approaches. MAC-level switching does not suffer from transmitting signals into incoming receptions, Bluetooth polling or operating system latency.

Conclusion
The interference and co-existence issues that exist with co-located Bluetooth and WLAN devices are amongst the most difficult in radio design because of the low power, cost and size constraints. Since neither standard contains a method for coordination, a mix of RF design practices and proprietary coordination mechanisms are used to ensure interoperability with negligible performance impact to the end-user.

musang_pulut

References

• Bluetooth SIG Core Specification version 1.2, 05 November 2003.
• S. Shellhammer, Packet error rate of an IEEE 802.11 WLAN in the presence of Bluetooth, IEEE P802.15 Working Group Contribution, IEEE P802.15-00/133r0, Seattle, WA (May 2000).
• O. Bar-Shalom, G. Chinn, K. Fleming, U. Gadammsetty, On the Union of WPAN and WLAN in Mobile Computers and Hand-Held Devices, INTEL Technology Journal, Volume 7, Issue 3, August 2003.
• Bluetooth and WiFi Coexistence Overview White Paper, 2003, www.siliconwave.com.
• How 802.11b/g WLAN and Bluetooth Can Play Together, White paper, 2004, Philips Semiconductor.
  

chiKenliTeL

tada gambar ka ~~~