Pressure Mounts in Next-Gen Mobile Phone Designs

Bluetooth uses a frequency hopping spread spectrum (FHSS) system in which the transmission band hops over 79 pre-defined 1 MHz channels. The hopping rate is roughly 1600 hops per second over a random pattern (note: adaptive frequency hopping will be deployed in the future). In this way, Bluetooth spreads energy over the entire band. However, since Bluetooth doesn't monitor the band before transmitting, it can easily interfere with other systems trying to use the same band. In this fashion, if 802.11 is transmitting or receiving when Bluetooth begins transmission, both air interfaces can fail to operate properly.

In contrast to Bluetooth, 802.11 does monitor its transmission band for other traffic before beginning to transmit. 802.11 employs direct frequency spread spectrum (DFSS) and orthogonal frequency division multiplexing (OFDM) air interfaces, and occupies roughly a quarter of the 83.5MHz ISM band. Since 802.11 will sense Bluetooth activity and not transmit if Bluetooth is active, WLAN service will be very seriously affected when Bluetooth interferes.

Fortunately, there are solutions to this interoperability challenge. Non-collaborative coexistence solutions include such techniques as adaptive fragmentation, which optimally adjusts packet sizes to minimize collisions, and adaptive frequency hopping, in which the Bluetooth interface selectively hops to channels where it can successfully transmit. Collaborative coexistence solutions involve communication between the 802.11 and Bluetooth interfaces in a handset and work to maximize the throughput of both interfaces while taking account of any quality of service (QoS) demands. Since Bluetooth is often used for voice communication, QoS is especially important, as the user will certainly notice if voice is interrupted.

Collaborative coexistence solutions can be both distributed, where each interface takes account of information from the other in determining when to transmit, or centralized, in which one of the interfaces assumes control and the second only transmits when it is signaled to do so. For example, the 802.11 media access controller (MAC) might assume control, taking account of any Bluetooth QoS demands and data transmission requirements, in addition to its own needs and optimally controlling both interfaces to make the best possible use of the transmission channel.

The Road to Digital Radios
Coexistence issues aside, form-factor limitations in modern handsets make the addition of new technologies that include a radio interface, such as Bluetooth and WLAN, very difficult. Normally a radio interface includes a transceiver function, power amplifier, and passive filtering and matching components. Consequently, the radio function is relatively large, making it very hard to meet form factor requirements in a small handset. Since integration is one of the most powerful tools available to semiconductor designers in reducing power and area, it is only nature to turn to integration to overcome the radio form factor challenge.

Of course, several alternatives exist for radio integration in a handset including multiple air interfaces. It might seem most straightforward to simply integrate all the radios together into a single radio IC and put all baseband processing functions in another. In this way, both the radio and baseband functions can be in an optimized process technology and very little new technology development is needed.

However, this approach suffers serious drawbacks. Clearly, there is no possibility to mix and match handset options in a modular fashion since every handset would have to include all incorporated air interfaces. Additionally, lack of processing capability on the radio IC would mean that aggressive control of the RF function by the processing logic would be lost, leading to lower yields and performance enabled by other possible options.

A more attractive approach is to simply integrate the RF and baseband functions of each air interface on a separate IC in deep submicron CMOS. In this way, each function is fully self-contained, allowing phone models to be developed with any number of desired radio functions to be included. Such a modular approach at the system level is especially attractive when many phone models must be designed under tight development time constraints. Additionally, this approach offers the benefit that the RF function is moved into the realm of deep submicron CMOS processing.

As opposed to designing the radio in a BiCMOS or SiGe technology, building the radio in advanced CMOS ensures that the radio function will be implemented in the most aggressively scaled technology the semiconductor industry has to offer. In addition to the other benefits of CMOS, this move to the most advanced CMOS processes available ensures that the radio will offer the lowest possible power consumption level and very low cost. Finally, tight coupling of the radio function with the baseband processing elements allows the radio to be automatically tuned, calibrated and tested by the logic functions. Higher production yields and better performance result.

Of course, integration of the radio function into CMOS with the baseband processing functions does require that a new radio architecture be implemented. Figure 3, shows a digital RF processor (DRP) that is suitable for CMOS integration. While the direct conversion receivers used in many of today's handset designs have served the industry well and literally billions of them have been shipped, their requirement for many stages of analog processing and high performance passive elements makes them unattractive for implementation in advanced digital CMOS. In contrast, the DRP receiver makes extensive use of sampled data processing and digital techniques, leveraging the very high logic density and fast clocking capability of advanced CMOS.