Hardware and software approaches implement multiple radios

Today’s on-the-go consumers may think they are carrying a single cell phone, but they’re actually carrying seven or more radios that handle multiband-cellular operation, Wi-Fi networking, Bluetooth connectivity, assisted-GPS (global-positioning-system) functions, and more. In the future, they might have radios that implement ad hoc device-to-device communications, in which one device takes advantage of a nearby device that has better coverage (see sidebar “Ad hoc sharing could mean more radios”). To fit all the necessary radio functions into a portable wireless device, researchers throughout 2008 have been developing miniature multifunction modules and exploring flexible, low-power SDR (software-defined-radio) architectures.

The traditional way of implementing multiple radios in a single product is simply to design in multiple discrete radios—a transmit/receive function for each communications standard your product aims to support. That approach can take up a lot of real estate, a problem that Epcos addresses with its miniature front-end LTCC (low-temperature-cofired-ceramic)-radio modules. “We ship … a fully tested RF system in a single package,” says Patric Heide, PhD, the company’s director of product development for modules. “In a mobile phone, you have the CMOS-based radio IC built in on one side, and, on the other side, you have the antenna. We provide a fully featured front-end module that goes between the two.” To shrink multiple-radio implementations, Epcos offers multiradio modules, such as a Wi-Fi/WiMax (worldwide-interoperability-for-microwave-access) module (Figure 1). See Reference 1 for a discussion on how to test the modules.

Modules such at the Epcos LTCCs can accommodate semiconductor components such as GaAs (gallium-arsenide) power amplifiers, but the CMOS-radio chips that accompany the front-end modules in the transmitter/receiver-signal chain embody many of the radio functions. You can use multiple CMOS chips to implement multiple radio standards. A more elegant approach involves using one flexible CMOS chip to implement multiple radio standards.

Justin R Rattner, vice president and chief technology officer of Intel, addressed that topic last June at the Design Automation Conference in Anaheim, CA . In a keynote speech, “EDA for digital, programmable multi-radios,” he noted that consumers want to “carry small yet live large”—that is, they expect compact designs to have multiple advanced features. To support this carry-small/live-large lifestyle, consumer devices must support “anywhere/anytime” collaboration, with the device able to communicate no matter what air interface—3G (third-generation) cellular, Wi-Fi, or WiMax, for example—is available.

Rattner acknowledged that we live in an analog world. “Analog is how we interact with the real world, but the technology favors digital,” he said. “So, we have to bridge the gap between the analog world and the digital world.” Bridging this gap will be important for the carry-small/live-large lifestyle. He further acknowledged that the amount of analog circuitry is increasing: In 2006, more than 70% of SOCs (systems on chips) had analog elements. Unfortunately, however, traditional analog development is not getting any easier in the face of transistor scaling. The reasons for this difficulty are increased mask costs; leakage and process-variation effects; output impedance; increased flicker noise; and reduced supply voltages, which decrease dynamic range.

His inclination is to employ “digitally assisted analog”—turning an analog problem into a digital one by relying on digital gates to improve analog performance. The technique exploits the computational nature of radio, which information theory embodies, to dramatically simplify radio architectures; to allow one radio to simultaneously act as many to, for example, facilitate seamless handoff between 3G and Wi-Fi networks; and to investigate the radio environment to see whether higher-performance networks are available.

In Rattner’s version of a digital-multiradio implementation, the traditional analog-receiver-signal chain, comprising a front-end module, a mixer, and a channel-select filter, for example, gives way to a simplified digital approach. Single chips can now implement a Wi-Fi or 3G radio. Tomorrow will bring integrated, programmable multiradio implementations to deal with any air interface they encounter. According to Rattner, Intel will probably need a year or so to produce commercial versions of such a digital radio, but the company has built a digital power amplifier in 65-nm CMOS and a fractional-N synthesizer in 90-nm CMOS.

An organization that’s far along in this effort is IMEC, which, at its Annual Research Review Meeting on Oct 14, 2008, in Brussels, Belgium, demonstrated an SDR platform. Liesbet Van der Perre, scientific director of wireless research at IMEC, says that the organization’s SDR prototype integrates key components for next-generation flexible mobile terminals. The prototype incorporates an RF-transceiver and programmable-baseband platform, enabling the measurement of the performance and power consumption in real-life conditions and for different operating modes, she says.

Van der Perre reports that IMEC’s flexible RF transceiver front end, SCALDIO (scalable radio chip), operates with all current and future cellular, WLAN (wireless-local-area-network), WPAN (wireless-personal-area-network), broadcast, and positioning standards in the 174-MHz to 6-GHz frequency range. IMEC’s programmable-baseband platform—BEAR (baseband engine for adaptive radio)—supports standards such as 802.11n, 802.16e, and mobile TV, and it is forward-compatible with the upcoming 3GPP-LTE (Third Generation Partnership Project/Long-Term Evolution communication) standard.

The connectivity-centric SDR platform makes flexible and efficient use of network and spectrum resources across heterogeneous environments. SDR also serves as an enabling technology for spectrum-centric, opportunistic, cognitive-radio applications. A cognitive radio can autonomously change its parameters based on interaction with the environment in which it operates, and it coexists with and uses the same spectrum resources as other wireless systems without introducing significant levels of interference. Full cognitive radio won’t be available until 2025 or 2030, Van der Perre estimates, but evolutionary cognitive-radio features—that SDR implementations enable—will become available in the interim.

As for SDRs, companies dedicated to the concept include Vanu, which applies the technology to base stations to enable them to simultaneously operate GSM (global-system-for-mobile) communications, CDMA (code division/multiple access), and Motorola’s IDEN (integrated digital enhanced network), for example. Another is BitWave Semiconductor, whose engineers have developed an SDR IC that can morph itself to work with at least 16 wireless-network interfaces, including GSM, WCDMA (wide CDMA), Wi-Fi, WiMax, and UMTS (universal-mobile-telecommunication-system) LTE. BitWave in early 2007 released the first prototypes of the device, the Softransceiver RFIC (radio-frequency integrated circuit). The IC is now nearing commercial release, with engineers testing the devices and characterizing them in production volumes

IMEC announced in October that Toshiba had licensed IMEC technology applicable to SDRs. And IMEC and Panasonic last month signed a joint-research contract concerning technologies in the semiconductor, networks, wireless, and biomedical fields. Under the terms of the contract, research will take place at IMEC’s Leuven, Belgium, facilities and research unit at Holst Centre in Eindhoven, the Netherlands. Part of the research will focus on dynamically reconfigurable SDRs.

John Irza, technical-marketing manager at The MathWorks, speaks from his company’s vantage point as a supplier of baseband- and RF-simulation tools for chip and system vendors. He estimates that bandwidth users employ only 15 to 20% of available spectrum at once, leaving plenty of room for a cognitive radio to operate. Cognitive radio must, however, be agile enough to defer to the primary user of the spectrum when that user wants access to the spectrum that cognitive radio has appropriated. Irza notes that cognitive radio’s challenges are both political and technical, requiring the approval of regulating bodies such as the Federal Communications Commission in the United States. Thus far, the FCC has supported the cognitive-radio concept.

As for the future, Irza sees a trend toward integrated baseband/upband devices that support BluetoothWi-Fi, GPS, and LTE communications standards as vendors overcome the challenges of moving digital-signal-processing functions closer to the antenna. IMEC’s Van der Perre cites figures from researcher ARCchart, which optimistically predicts that manufacturers will ship 157 million SDR-enabled handsets—or 11% of all handsets—in 2011. The researcher’s pessimistic forecast is 74 million SDR-enabled handsets for the same year.

Van der Perre doesn’t cite figures for SDRs or cognitive radios beyond 2011, but she does predict that radios’ ADCs will move ever closer to their antennas, yielding an intuitive, flexible full cognitive radio by approximately 2030. As a step toward that goal, IMEC presented at the IEEE Asian Solid-State Circuit Conference in November in Fukuoka, Japan, a 2.4-GHz sigma-delta ADC it fabricated in 90-nm CMOS. Research at IMEC continues on a scalable energy-efficient spectrum-sensing engine that could bring cognitive radio yet another step closer.