You are on page 1of 7

Multi-Radio Coexistence: Challenges and Opportunities

Jing Zhu, Alan Waltho, Xue Yang, and Xingang Guo Communication Technology Lab, Intel Corporation, Hillsboro, OR 97124, USA {jing.z.zhu, alan.e.waltho, xue.yang, xingang.guo}@intel.com
Abstract in this paper, we study the emerging heterogeneous multi-radio network, focusing on interference issues due to simultaneous operation of multiple radios. A comprehensive review is given on the sources of coexistence interference, including transmitter noise, receiver blocking, and intermodulation. We compare various coexistence techniques, and relevant standardization efforts. Then, we propose a media independent coexistence service (MICE) layer, and suggest that the performance of a multi-radio system can be further improved if necessary support and modification is added to individual wireless technology, e.g. air interface / wireless protocols. Specifically, we show that the proposed coexistence-aware TXOP adaptation achieves up to 50% improvement of the transmission efficiency in a WiFi/Bluetooth dual-radio device, where the 802.15.2 PTA (packet traffic arbitration) technique is used for coexistence. Key words: coexistence, standardization, and air-interface I. INTRODUCTION A communication and computing device, such as cellular phone, PDA, laptop, is being equipped with more and more radios. It is certainly desirable from the users perspective (see Fig.1) to allow all the radios to simultaneously operate without interfering with each other. One such example is todays cellular phone, which is widely used with Bluetooth (BT) headsets. Furthermore, WLAN radios are being integrated into cellular phones, causing more interference in the 2.4GHz ISM band and the adjacent bands.
case. On the other hand, proximity is much more difficult to deal with, because there no interface or common communication channel between heterogeneous devices, e.g., WiFi and Bluetooth. The proximity problem is very common in real world scenarios in which, users are using different wireless technologies (e.g. WPAN, WLAN, WWAN, etc.) in the same room. Below, we list some of the typical collocation usage cases: Voice Call with Bluetooth headset: a mobile device is relaying a Voice call via WLAN or WWAN connection and from a Bluetooth headset. Seamless Vertical Handover: a mobile device running network applications (VoIP, net-meeting, video streaming) is roaming from inside (covered by WLAN) of a building to outside (covered by WWAN) with seamless vertical handover. Wireless Peripheral and Internet Access: computer peripherals are going wireless, such as (wireless) USB, (BT) keyboard / mouse, and others. A client (PC) is connecting to Internet via WLAN or WWAN, and the peripherals are connecting the client (PC) via WPAN or WLAN. Wireless Gateway: a residential gateway that have both radios, one for in-bound (e.g. WLAN), and the other for out-bound (e.g. WWAN) Wireless Display and Internet Access: a mobile device uses WLAN to display images on a remote display while receiving such images from the Internet via WWAN Table 1 shows the frequency bands of the common radios that are likely to be integrated in a single device. We categorize them into: WAN, LAN, PAN, location service, and digital broadcast. Clearly, lots of radios are operated around 2GHz. Therefore, these radios (e.g. WiMAX/UMTS, WiFi, and Bluetooth) are likely to interfere with each other. Fig.2 shows the spectrum of (unlicensed, 2.4GHz) Bluetooth and (licensed, 2.5GHz) WiMAX, and the minimum separation is about 30 MHz. In addition a UWB device can also cause or receive interference from other radios across the frequency band 3-10 GHz. In this paper, we will first briefly describe the causes of coexistence interference, and demonstrate through simple examples the level of coexistence interference that could exist given the small coupling loss between radios (Section II). Then, we will compare various coexistence techniques that are proposed in the literature and by industry (Section III). We will also discuss the ongoing IEEE standards related to coexistence (Section IV), and show how to improve the system performance of a multi-radio device by making airinterface adapt to coexistence interference (Section V). Finally, we conclude the paper (Section VI).

Fig.1 Multi-Radio Network

Without loss of generality, there are two coexistence scenarios: proximity and collocation. Proximity indicates the case where devices are not in the same platform but are close enough so that they will interfere with each other if operated simultaneously. The coupling of the interference source to the victim is by radiation, primarily between the antennas. Collocation indicates the case where multiple radios are in the same physical unit so that mutual interference can be caused by conduction as well as radiation. The (over-the-air) free space propagation provides greater isolation in the proximity case and interference is less severe than in the collocation

1095-2055/07/$25.00 2007 IEEE

Table 1: Band and Power of Wireless Technologies

Usage
WAN

Radio
Cellular (GSM, CDMA, GPRS)

Band

824 - 894 MHz, 880 960 MHz 1770-1880 MHz, 1850-1990 MHz UMTS (W-CDMA, 1920 -1980 MHz CDMA 2000, 1XEV- 2110 -2170 MHz DV, 1XEV-DO)
802.16 (WiMAX)

Transmit Power 30dBm

30dBm

LAN
PAN

802.11 (WiFi)
802.15.1 (Bluetooth)

2.300 2.400 GHz 22dBm 2.496 2.690 GHz 3.300 3.800 GHz 2.412-2.4835 GHz 16dBm 4.9 5.9 GHz
2.4GHz 10dBm
-41 dBm/MHz RX only

UWB (Wireless USB) 3.1 10.6GHz


Location GPS Service Digital DVB-H TV Broadcast

1575.42 MHz 1227.6 MHz. 1.6 1.7 GHz in US RX only 470 to 820 MHz in Europe

However, the PL model is only effective in defining RF signal propagation in the far field. If two radios are collocated on the same device, and the distance is very small (e.g. near field), we have to also consider antenna spatial configuration, mounting structure, and so on. Furthermore, in the collocation case, coupling can occur by both radiation and conduction so that the coupling loss is much more difficult to model, and varies greatly depending on the characteristics and material of antenna and other components on the device. For example, that the coupling loss between BT (Bluetooth) and UMTS (ultra mobile telecommunications system) radios on a cell phone can vary between 10 and 30 dB depending on the specific design. Co-channel or in-band interference is the most common interference. However, the CCI among homogeneous devices has been handled within individual wireless technology through various media access mechanisms, such as TDMA, FDMA, or CDMA. Our focus in this paper is on interference among heterogeneous devices where the interference is due to out of band mechanisms. Fig.3 shows a typical (collocated) coexistence interference scenario. Radio A.1 (receiver) is being interfered by Radio A.2 (transmitter), where there are three possible interference mechanisms receiver blocking, transmitter noise and inter-modulation.

30MHz

BT (unlicensed)

WiMAX (licensed)

Fig. 2 WiMAX and BT Spectrum Occupation

Fig.3 Coexistence Interference between Collocated Transceivers

II. COEXISTENCE INTERFERENCE A. Fundamentals It is well known that as an RF (radio-frequency) signal is propagated, its strength degrades. A path loss (PL) model, describing signal strength as the function of distance, has been widely used to characterize such behavior. The free-space PL model is the simplest model and is mainly applicable to point to point communications in an outdoor environment where LOS (Line-Of-Sight) exists between transmitter and receiver. Other models, that incorporate additional looses due to shadowing and refraction are applicable to most mobile scenarios. The free space model is modified for indoor communications. Eq.(1) shows an empirical path loss model [2] for indoor and 2.4GHz ISM band. 40.0 20 log(d ), ( d 8) (1) Lp = 58.5 33 log(d / 8), (d > 8) where Lp is path loss (in dB), and d is link distance (in meters).

Receiver blocking is mainly due to the limited dynamic range of power amplifier and analog-to-digital (A/D) converter. For example, a typical WLAN receiver has a blocking limit of -30 dBm at 2GHz. If the total input power exceeds this limit, the received signal strength of the desired signal will suffer. Two parameters are used to model receiver blocking in [3]: P3dB and k. The signal degradation (in dB) due to receiver blocking is written ([3]) as (2) B = max{0, (Pin P3dB)k +3 } where Pin indicates the total power level that are put into receiver. P3dB is also referred as the 3dB compression point, which will decrease as the spectral separation increases. k is the dB value of the degradation of the received signal strength for every dB increase of input power. A transmitter will also generate out-of-band (OOB) emission (IOOB) due to imperfect filtering, or phase / LO (local oscillators) noise. Such wideband noise is also called transmitter noise, and can fall within the receiving band of another receiver that is in close proximity or collocated.

Inter-modulation (IIM) is caused by the non-linearity of components in the signal path, e.g. LNA (low noise amplifier). Given two signals with their carrier frequencies at f1 and f2 (f1 < f2), respectively, the 3rd order IM interference appears at f1 (f2 f1) and f2 + (f2 f1), and its power level (dBm) can be written as the function of the power of input signals (denoted as P(f1) and P(f2)), i.e., IIM (f1 (f2 f1)) = 2P(f1) + P(f2) 2IP3 + 10log10(f) (3) IIM (f2 + (f2 f1)) = 2P(f2) + P(f1) 2IP3 + 10log10(f) (4) IP3 is called the third order IM intercept, and can be explained as If P(f1) = P(f2) = IP3 and f = 1 (5) IIM (f2 + (f2 f1)) = IIM (f1 (f2 f1)) = IP3 Notice that we use f to denote the overlapping ratio. When the separation between two signals (f2 f1) is less than the bandwidth of the signal, the 3rd IM interference will overlap with the band of the desired signal, and f indicates the ratio of this portion to the whole IM bandwidth. f = 0 means there is no overlapping so that IM is negligible. Now, we can summarize the impact of coexistence interference as the following SINR form:

signal. Table 3 shows the ACS requirement for WiMAX TDD mobile device. Assuming a collocated WiFi transmitter operated at channel 11 and transmit power 16dBm, the ACS of the WiMAX receiver at channel 11 is 66dB according to Table 3. Hence, the coexistence interference is 50dBm. If we want to maintain the coexistence interference below the noise floor (-110dBm), the required MCL is 60 dB.
Table 4: OOB of WiFi transmitter (Tx Power: 16dBm, BW:20MHz )

Frequency Spacing OOB power density (by spectrum mask) 6dB margin

20MHz -85 dBm/Hz


-91 dBm/Hz

30MHz. -97 dBm/Hz


-103dBm/Hz

Table 4 shows the OOB emission level of a typical IEEE WiFi transmitter. It is specified in IEEE 802.11 a/g that the OOB power density must be lower than -40dB at 30MHz separation, and -28dB at 20MHz separation. Considering 16dBm transmit power and 20MHz bandwidth (-57dBm/Hz), the OOB power density must be lower than -97dBm/Hz at 30MHz separation. Commercial products can achieve about 6dB better than the requirement, leading to -103dBm/Hz. Considering 40dB coupling loss between a WiFi transmitter and a WiMAX S / 10 B / 10 (6) receiver and 30MHz frequency spacing, the OOB power SNIR = B / 10 N 0 + ( I OOB + I IM ) / 10 density at the WiMAX receiver is therefore -143 dBm/Hz. Wherein S is received signal strength without coexistence Additional 27dB isolation must be provided to bring it down interference, and N0 is background noise. Both IOOB and IIM are to the noise floor -170dBm/Hz. Better transmitter design can considered only for the energy that fall within the band of the lower this figure. It was shown in [6] that the OOB of a desired signal, i.e. co-channel interference. While, receiver 2.4GHz WLAN transmitter can be as low as < -170dBm/Hz at blocking (B) should be considered for a given spectral 300MHz separation, which however will be extremely separation between interference and desired signals. difficult to achieve at 30MHz separation. B. Coexistence Analysis 3) Inter-Modulation (WiMAX TX and WiFi RX) In this subsection, we will use simple examples to demonstrate Intermodulation between the collocated WiMAX transmitter how much coexistence interference is among WiMAX / and a remote source can occur within the receivers LNA or UMTS, BT, and WiFi radios. mixer to generate on channel interference. Given WiMAX 1) Receiver Blocking (UMTS TX and BT RX) transmit power of 22dBm and coupling loss of 40dB, the input Table 2: Receiving blocking of Bluetooth radio power from WiMAX transmitter to WiFi receiver is -18dBm. Frequency 30MHz 2GHz 2.484 3 GHz Lets use P to denote the WiFi receiving power. Table 5 shows 2GHz 2.399GHz GHz 3 12.75 GHz GHz IIM (based on Eq.(3)).
P3dB
-10dBm -27dBm -27dBm -10dBm
Table 5: IIM (dBm) from WiMAX Tx to WiFi Rx

Table 2 shows the P3dB requirement for a BT radio in different frequency bands as specified in [5]. If we consider the 2-2.399 GHz band, where a UMTS device could operate, the P3dB requirement is -27dBm. As shown in Table 1, the transmit power of a UMTS device can be as high as 30dBm. As a result, a minimal coupling loss (MCL) of 57dB between a UMTS transmitter and a Bluetooth receiver is required to prevent blocking. However, the coupling loss between radios on a cell phone is typically between 10 and 30 dB therefore severe blocking will occur unless additional measures are taken. 2) Transmitter Noise (WiFi TX and WiMax RX)
Table 3: 2.5GHz WiMAX TDD Mobile Device ACS Requirement [17] Frequency Separation (MHz) 5 10 15 20 ACS (adjacent channel selection) (dB) 40 59 66 66

P = -30dBm P = -60dBm P = -80dBm

IP3= -30dBm -24 -84 -124

IP3 = -20dBm -44 -104 -144

IP3= 10dBm -64 -124 -164

IP3= 0dBm -84 -144 -184

Adjacent channel selectivity (ACS) is another parameter to measure the transmitter noise at receiver side. It measures how much rejection a receiver can maintain at a given frequency separation relevant to the center frequency of the desired

A channel-bonding technique has been adopted by the next generation WLAN standard (i.e., 802.11n), to increase link peak data rate. As a result, the WiFi signal could occupy 40MHz so that when intermodulation with a single narrow band source occurs the 3rd order IM interference will be 80 MHz wide. As shown in Fig. 2, the separation between the edges of the WiFi and WiMAX spectrums can be as small as about 20MHz. As a result, the 3rd order IM interference will cover the lower 20MHz portion of the fat channel that combines channel 6 and 11, i.e., f =10log10(20/80)=-6.

III. COEXISTENCE INTERFERENCE MITIGATION Table 6 summarizes the existing coexistence interference mitigation techniques, which fall into two categories: physical technique (PHY), and media access control (MAC).
Table 6: Comparison of Coexistence Interference Mitigation Techniques

Technique

Issues

PHY: static spectrum masking sacrifice antenna isolation performance (e.g. shielding filter reduces filtering sensitivity) beam-forming (BF) additional cost and interference cancellation (IC) size MAC: suboptimal without dynamic frequency selection (DFS) air-interface support transmission power control (TPC) control overhead time sharing (TS)

idea is to remove interference from the input by either spatial diversity (e.g., pointing a null to the interferer) or signal processing. Notice that they are used only for external (overthe-air) interference. However, the PHY solutions have to be designed on a case by case basis. They usually come with additional hardware requirement, and are limited by device size. Furthermore, coexistence interference only occurs when multiple radios are simultaneously operated. There are still many cases where only one radio is running at the time. However, many PHY solutions are embedded in hardware, and therefore they are not easy to modify. They may even negatively impact the performance (e.g. insert loss) when the device is working in a single-radio mode. B. Media Access Control (MAC) A MAC approach is reactive, and provides protection only when it is necessary. The most common methods in this category are dynamic frequency selection (DFS), transmission power control (TPC), and time sharing (TS). The basic idea is to detect and then avoid interferences by adjusting frequency, transmit power, or timing. DFS allows a radio to dynamically select a channel with the least interference. For example, AFH (adaptive frequency hopping) defined in IEEE 802.15.2 [1] avoids WiFi interference by excluding from its hop set those frequencies that lie within the band of the active WiFi channels. For collocated radios AFH may be implemented by means of a direct interface between the radios, elsewhere some form of energy detection or error rate test is required. The energy detection measures the RSSI (received signal strength indicator), and the error rate test measures the PER (packet error rate). However, the main limitation to DFS is that the choice of operating frequency is not always available to the victim receiver. For example a WiFi client is tied to the frequency of the access point with which it is operating. Moreover, DFS is unlikely to be useful if interference covers a wideband (e.g. receiver blocking or transmitter noise). TPC (transmission power control) forces a radio to operate at its lowest transmit power subject to link budget measurement or calculation. As a result, when a client is closer to Access Point or Base Station, it will transmit less power, and cause less interference to other coexisting radios. Moreover, a receiver that is suffering from coexistence interference can notify its peer transmitter to increase transmit power. But TPC can not help if client is at cell edge or transmit power is already at its maximum. TS (time sharing) is also used by the industry to support coexistence, particularly for Bluetooth and WiFi. The basic idea is to schedule multiple radios in time domain so that they do not overlap with each other, e.g., alternative wireless medium access (AWMA) and packet traffic arbitration (PTA) (see Fig.4). AWMA is a static scheduling method, which divides an IEEE 802.11 beacon interval into two subintervals, one for WLAN and the other for WPAN. The subinterval duration is managed by the MSE (medium sharing element) in the IEEE 802.11 beacon, and therefore AP support is mandatory for AWMA. Unlike AWMA, PTA does not need any AP support, and work on a per packet basis. A radio will make a request to a PTA controller before each transmission.

A. Physical Technique (PHY) The transmitter spectrum mask is defined by regulatory bodies, such as FCC, and all transmitters are required to comply with the relevant mask. It is intended to protect other spectrum users when physically separated from the transmitter by reasonable distances. As specified in FCC Part 15. [8], the emission limit of a class A1 digital device is defined at a distance of 3 or 10 meters. These regulatory spectrum masks are neither intended nor sufficient to prevent interferences between radios that are collocated or in very close proximity (say < 1m). It was shown in the previous section that > 40dB isolation is required for Bluetooth and UMTS coexistence. However, achieving >40dB isolation in a small form-factor device, such as mobile phone, is not easy to satisfy. On the other hand, it will increase cost and power consumption to improve the spectrum masks to meet the requirement of multiradio coexistence. Increasing antenna to antenna coupling loss by careful antenna placement is another common method to mitigate coexistence interference. Spatial placement or orientation of antennas can increase the coupling loss between antennas. Shielding will add further isolation between the radios circuitries, and usually is provided at the package or platform level. It was shown in [9] that it was possible to encapsulate a shield within the die. Band pass or low pass RF filtering at the transmitter output is often used to reduce harmonic emissions. This same filter also reduces OOB emissions occurring at large frequency separations but, due to the low Q factor available with current technology, they have little effect on emissions close to the transmit frequency. A typical Ceramic RF filter provides about 15 dB of attenuation not be useful at frequency separations of less than 100 MHz when operating at 2.5 GHz. Alternatively when the frequencies are known, a notch filter can be used to null the spurious, harmonic, or inter-modulation interference. Beam-forming (BF) and interference cancellation (IC) can also be used to mitigate coexistence interference. The basic
1 Class A digital device. a digital device that is marketed for use in a commercial, industrial or business environment, exclusive of a device which is marked for use by the general public or is intended to be used in the home, page 6, FCC Part 15 [9]

If the PTA controller detects a potential collision (with the other collocated radio), it can reject or admit the request based on a predefined policy. For example, a priority policy is defined in [1] for WLAN and WPAN coexistence an IEEE 802.15.1 SCO packet should have a higher priority than IEEE 802.11b DATA MPDUs and an IEEE 802.11b ACK MPDU should have a higher priority than all IEEE 802.15.1 packets. However, a TS method is not useful if a radio uses continuous wave form, and has 100% time occupancy during the connection.

providing a unified and scalable multi-radio coexistence support.


Standard
802.15.2

Table 7: IEEE Coexistence relevant Standards Publish Scope Year


2003 recommended practice for coexistence of IEEE Std 802.15.1-2002 WPANs and IEEE Std 802.11b-1999 in Unlicensed Frequency Bands

802.16.2

802.11h

Fig. 4 PTA and AWMA in 802.15.2 [1]

802.16h

The MAC coexistence solutions usually need support from air-interface and wireless protocols, and have control overhead. A Common Spectrum Coordination Channel (CSCC) etiquette protocol is proposed in [7] to exchange information among radios over the air for the purpose of channel selection and power adaptation. IV. COEXISTENCE STANDARDIZATION Table 7 lists the IEEE standards related to coexistence. Whereas, 802.19 and P1900.2 are focused on analysis and evaluation methodology; 802.15.2, 802.16.2, 802.11h, and 802.16h aim at providing solutions or facilitations to enable coexistence. Both 802.16h and 802.11h use DFS to avoid interference by adaptively selecting a channel with less interference. 802.11h further uses TPC to minimize transmit power based on link budget calculation or estimation. Unlike 802.11h and 802.16h, which only consider the proximity case, 802.15.2 considers both proximity and collocation, and defines the following techniques: alternating wireless medium access (TS) packet traffic arbitration (TS) deterministic interference suppression (IC) adaptive interference suppression (IC) adaptive frequency hopping (DFS) However, almost all todays IEEE standards related to coexistence are media dependent, and targeted at mitigating co-channel interference that comes from other in-band devices. There is relatively lack of unified approach to measure, detect, and avoid coexistence interference from outof-band (e.g. receiver blocking, transmitter noise, and intermodulation interference). For example, if there are 6 radios on a platform, we need to consider 26 7 (=57) interference scenarios (that has more than one radios). Hence, increasingly we need to standardize a media-independent coexistence service layer, integrating various coexistence techniques and

802.19

P1900.2

recommended practice for coexistence between multipoint (MP) systems and pointto-point (PTP) systems in the 10-66GHz, and among FBWA (fixed broadband wireless access) in the licensed 2-11GHz bands 2003 specific requirements for mechanisms for dynamic frequency selection (DFS) and transmit power control (TPC) that may be used to satisfy regulatory requirements for operation in the 5 GHz band in Europe ongoing amendment for improved mechanisms, policies and medium access control enhancements, to enable coexistence among license-exempt 802.16 systems, and to facilitate the coexistence of such systems with primary users ongoing recommended practice for metrics and methods for assessing coexistence of IEEE 802 wireless networks ongoing technical guidelines for analyzing the potential for coexistence or in contrast interference between radio systems operating in the same frequency band or between different frequency bands.

2004

Applications

Traffic
M

Local Policy t Multi-Radio Management

Multi-Radio Coexistence Coordinator


Information Service
Coexistenc e Policy
Radio Control

WLAN
GPS

WPAN

WWAN
DVB-H

Protocol and Device Hardware


Fig.5 Media Independent CoExistence (MICE) Service Layer

Fig. 5 shows the future multi-radio system with the media independent coexistence service (MICE) layer with the key elements discussed in the following: Information Service: to gather necessary information for coexistence interference analysis and guiding radio control. For example, we can collect time schedule (e.g., start time and end time of next transmission or reception),

Transmission Efficiency

spectrum / energy profile (e.g., channel width, central frequency, modulation, power), transceiver RF profile (e.g., P3dB, IP3, etc.). Notice that PHY coexistence techniques are still provided by each radio vendor, since they have to be built into individual radio. However, it will be helpful to expose the capability and the profile of these techniques to the MICE through a standardized interface. As a complementary step, a MAC technique can be triggered whenever the PHY solution is not sufficient to suppress coexistence interference. Radio Control: to provide commands for coexistence interference mitigation techniques, e.g., scheduling multiple radios in frequency, power, and time. Coexistence Policy: to provide priority and policy when collision or conflict occurs among radios. Other than a unified coexistence service layer, it is also necessary to enhance air-interface / wireless protocol, and make them flexible enough to adapt to coexistence interference. Next we will use a simple example to show why such support is important. V. COEXISTENCE-AWARE TXOP ADAPTATION Fig.4 shows a device with WLAN and WPAN radios simultaneously operated. Collision occurs whenever there is temporal overlapping between two. The notations in the figure are explained as follows: T: current time X: starting time of the next (or ongoing) WPAN activity (TX or RX) Y: ending time of the next (or ongoing) WPAN activity (TX or RX) L: WLAN TXOP (transmission opportunity) duration, Wherein, TXOP is defined as the time from the starting of a data transmission to the reception of the ACK frame.

Fig.7 BT HV3 link

Now, we use simulation to show how much benefit can be obtained from TXOP adaptation. We use transmission efficiency as performance metric, defined as the ratio of the time used for WiFi TX (from beginning of a transmission to the reception of ACK frame) to the total time. We assume a Bluetooth HV3 link with periodic traffic pattern: one TX followed by one RX every 6 slots, and BT slot time is 625s. Whenever there is temporal overlapping between WiFi TX with BT RX, the WiFi radio is forced to stop and reset the back-off counter.
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 200 150 100 50 1

Tmin (WiFi slots)

Fig.8 impact of the minimum TXOP on transmission efficiency

Fig.6 Collision of WLAN and WWAN radios on a multi-comm device

Normally, the timing of WPAN activity (TX or RX) is known in advance, and collision can be avoided by forcing WLAN radio to freeze its transmission like in PTA [1]. To improve WiFi channel utilization, it is better to adjust the TXOP duration (denoted as L) to the optimal value, i.e., X T.

For the WiFi radio, Contention Window (CW) is set to 32, back-off slot is 9s, and the maximum TXOP is set to 1.8ms. Also we assume there are no other WiFi stations competing for the channel, and we assume saturated traffic transmitted from the multi-radio device to access point. The idle interval between two consecutive Bluetooth RX slots is 3.125ms. Lets use Tmin to indicate the minimum TXOP duration that a WiFi TX can use. Every idle interval can hold up one transmission if Tmin=200 slots, leading to the transmission efficiency of 0.482, and two transmissions if Tmin=1 slot, leading to the transmission efficiency of 0.763. Fig.6 shows the simulation results on the transmission efficiency as the function of Tmin. We observe up to 50% improvement in transmission efficiency. It is easy to enable such TXOP adaptation if the multi-radio device is the transmitter. Because the transmitter initiates the transmission sequence, and can decide the best TXOP duration based on coexistence interference at the very beginning. However, if the multi-radio device is a receiver, we have to let receiver feedback the reduced TXOP duration to the transmitter when a future collision due to coexistence is predicted. This will help avoid interference from Bluetooth TX to WiFi RX. It works as follows. A multi-radio device receives a RTS frame from AP via its WLAN interface. If the current value of TXOP duration (indicated by the RTS frame) is too long so that the WiFi RX will collide with the TX activity of other
2 3

200 x 9 / (625 x 6) = 0.48 (625 x 5 32 / 2 x 2 ) / (625 x 6) = 0.76

radio (e.g. Bluetooth), the client will put a reduced value (determined by how much idle time is left before other radio starts their next transmission) in the TXOP duration field of the CTS frame, and mark the Power Management bit in the frame control field. It indicates that the client will be unavailable after conclusion of the current frame exchange. After receiving the CTS frame with the reduced value, the transmitter should finish its transmission earlier, according to the new value, and send out a CF (contention free)-end frame to inform its neighbors of the release of the channel. Furthermore, AP will not transmit any more frames to the client because of the Power Management bit. After the disruption finishes, the client can send out a data or null frame with the Power Management bit turned off to resume its activity. VI. CONCLUSION In this paper, we have shown that increasingly more and more (heterogeneous or homogenous) radios will be simultaneously operating in close proximity or even on a same physical device. Tomorrows mobile device will be equipped with radios for WPAN/WLAN/WWAN access, location service (GPS), and digital TV reception (DVB-H). However, the coexistence interference among these radios is becoming the bottleneck in both collocated and close proximity environments. Regulatory standards related to coexistence are inadequate to provide interference protection in these environments, a problem that becomes more acute as the form-factor continues shrinking, and use of the radios becomes more ubiquitous. We discussed three major sources for coexistence interference: transmitter noise, receiver blocking, and inter-modulation, and emphasized that they can occur among radios even when they are operated in different bands. We then compared various coexistence interference mitigation techniques. The PHY techniques are proactive and lack of flexibility, while the MAC ones are reactive but might introduce additional control overheads. We also reviewed the IEEE standard efforts related to coexistence, and proposed a unified media independent coexistence service layer to support scalable coexistence solution. Last, we use a simple example to demonstrate the benefit of bringing coexistence awareness into the design of air-interface and wireless protocol. Simulation result shows up to 50% performance improvement with the proposed solution. VII. ACKNOWLEDGEMENT We thank Adrian Stephens, Jose Puthenkulam, and Eldad Perahia from Intel for their valuable comments and suggestions on this work. We would also like to thank York Liu from Intel for providing Fig. 2. REFERENCE
[1] IEEE Std 802.15.2 , Coexistence of Wireless Personal Area Networks with Other Wireless Devices Operating in Unlicensed Frequency Bands [2] A. Kamerman, Coexistence between Bluetooth and IEEE 802.11 CCK Solutions to Avoid Mutual Interference, Lucent Technologies Bell Laboratories, Jan. 1999, also available as IEEE 802.11-00/162, July 2000. [3] M. Konrad, W. Koch, and J. Huschke, Coexistence Analysis of Bluetooth and Cellular UMTS in the 2500-2690 MHz Band, IEEE WCNC 2006.
TM

[4] S. Timiri, RF Interference Analysis for Collocated Systems, Microwave Journal, 1997. [5] Bluetooth SIG, Specification of the Bluetooth System, Bluetooth SIG, http://www.bluetooth.org, November 2003. [6] C. P. Huang, et al, A Compact High Rejection 2.4GHz WLAN Front-End Module Enables Multi-Radio Co-existence Up to 2.17 GHz, IEEE Radio Frequency Integrated Circuits (RFIC) Symposium, 11-13 June 2006. [7] X. Jing and D. Raychaudhuri, Spectrum Co-existence of IEEE 802.11b and 802.16a Networks using the CSCC Etiquette Protocol, IEEE 2005. [8] FCC Part 15. http://www.fcc.gov/oet/info/rules/part15/part15-8-14-06.pdf [9] M. Gynor, et al., System-in-Package for WLAN/PAN Aids Coexistence with Digital Cellular, pp. 30-40, January, 2003. [10] A. Waltho, Performance Analysis and Design Considerations for MultiRadio Platforms, Intel Developer Forum, 2006, Taipei. [11] IEEE Std 802.16 TM-2004, Air Interface for Fixed Broadband Wireless Access Systems [12] IEEE Std 802.16eTM-2005, Amendment2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands [13] IEEE Std 802.16.2 TM-2004, Coexistence of Fixed Broadband Wireless Access Systems [14] IEEE Std 802.11a-1999, High-Speed Physical Layer in the 5 GHz Band [15] IEEE Std 802.11g TM-2003, Amendment 4: Further Higher Data Rate Extension in the 2.4 GHz Band [16] IEEE Std 802.11h TM-2003, Amendment 5: Spectrum and Transmit Power Management Extensions in the 5 GHz band in Europe [17] Mason Communications Ltd Technical White Paper, 2500-2690MHz, 2010-2025MHz and 2290-2302MHz Spectrum Awards Engineering Study (Phase 2), Nov. 2006

You might also like