A block diagram for a typical software-defined radio receiver is shown in the figure. An SDR is designed to receive multiple RF carriers, with 15-20 MHz typical bandwidths, using one or more air standards. These carriers pass through a band-select filter, which reject blockers and interferers from adjacent frequency bands. A low-noise amplifier (LNA) follows, providing gain to the incoming signals. The analog RF mixer then down converts the desired signal to a convenient intermediate frequency (IF) for digitization. The down convert is followed by an integrated IF to baseband receiver subsystem. The AD6654, for example, incorporates a high performance 14-bit, 92-MSPS ADC to digitize the IF spectrum, and a four-channel digital down converter (DDC) that tunes and filters the desired signal. The output of the DDC consists of a channel-filtered digital IF signal, which is then demodulated by a TigerSharc DSP. With this architecture, the DDC and DSP can easily be reprogrammed to adapt to new cellular standards or changes in frequency spectrum allocation.

With the architecture for SDR established, cellular operators must consider where such a system can be deployed to provide the most benefit. For cellular infrastructure, one major consideration is the spectrum allocation among different regions of the world. Perhaps the strongest case for SDR deployment is in the US, which allows multiple air standards in the same frequency band-a result of many air standards being deployed, as well as consolidation among the different operators. A good example is the US 1900 MHz band. This band currently has IS-136, IS-95, GSM, CDMA2000 and UMTS (proposed). Deploying SDR would allow operators to use a single platform to process multiple air standards.

Unlike the US market, the European market has each standard allocated to an independent frequency band. Although this entails additional complexity, some advantages can be realized. GSM is deployed at 900 MHz and 1800 MHz; UMTS will be deployed at 1900-2100 MHz. Because of the way the frequency spectrum is allocated, a different band select filter, VCO and mixer would be needed for each frequency band. Still, board manufacturers could support a common radio platform instead of multiple platforms, only having to change the bill of materials to support the different bands.

In addition to understanding the frequency allocation for world cellular markets, operators must also understand the cost advantages and disadvantages of deploying single RF carrier systems versus multi-carrier software-defined radios. The component costs associated with a single RF carrier design have continually dropped as the 2G/2.5G markets have matured; but premium performance components are necessary in order to meet the stringent technical requirements of multi-carrier SDR platform. Comparing the direct component costs of single carrier designs versus multi-carrier designs is not realistic, however. The real benefit of multi-carrier systems is the ability to process more than one carrier without adding additional components. In secluded, lightly populated areas, single carrier deployment will make the most economic sense, while a multi-carrier system will be appropriate in areas with a moderate to large population and a higher demand for capacity. An SDR is designed to accept multiple RF carriers, so the cost per RF carrier will be reduced as more carriers are added. Thus, the operator must evaluate how many carriers must be utilized before the multi-carrier system becomes cost effective. The current cost of components puts this threshold at 4 to 6 carriers.

Another cost advantage for SDR is ease of migration. Several US operators, having already deployed IS-136, are now migrating to GSM/EDGE, and will ultimately migrate to UMTS. The costs associated with deploying each of these systems could be in the billions of dollars, but would be reduced substantially if the radio boards were left unchanged and only the software for the DDC and DSP needed to be changed. This would allow IS-136 to coexist with GSM/EDGE, so an operator could replace IS-136 carriers with GSM/EDGE carriers dynamically, meeting the demand for higher data rates through soft allocation.

The final consideration for developing SDR is the technical challenge it presents. The radio receiver is designed to accept a large bandwidth with multiple RF carriers, so it must absorb both desired and undesired signals while maintaining its performance. The analog-to-digital converter (ADC) traditionally limits the receiver performance. The figures of merit for an ADC are signal-to-noise ratio (SNR) and spurious-free dynamic-range (SFDR).

SNR can be expressed as the ratio of signal energy to noise energy in a spectrum. Using SNR, input voltage, and termination impedances, the noise figure (NF) can be computed. This, when combined with the other RF/IF components of the receiver will determine the overall sensitivity of the receiver. SFDR is defined as the ratio between the rms amplitude of a single tone and the rms amplitude of the worst spur as the tone is swept through the ADC input range. It is common for the worst spur to be harmonically related.

It is also important to understand the technical requirements of each cellular standard, and how they relate to these figures of merit. In Europe, for example, the GSM 900 MHz band presents the most challenging standard, as it requires the receiver to detect a signal as small as -104 dBm, or as large as -13 dBm, in the presence of blockers and interferers. This means that the ADC must maintain a 91-dB minimum SFDR. In reality, however, the GSM specification requires an additional 9-dB carrier-to-interferer ratio (C/I), so the true SFDR requirement is 100 dB. This represents the minimum requirement; most basestation manufacturers will strive to surpass this to offer a competitive advantage. With a wideband receiver, a large undesired signal cannot produce a harmonic or spur larger than 100 dB, as this spur could fall on top of the desired signal, resulting in a dropped call. In contrast, the US 1900 MHz and European 1800 MHz GSM markets relax the specifications so that the receiver only needs to detect signals from -23 dBm to -104 dBm, thus requiring only 90-dB SFDR including the 9-dB C/I specification.

A similar calculation can be performed for SNR. In order for a receiver to properly process an RF carrier, 6-dB SNR must be maintained through the receiver. For 900 MHz GSM, the ADC needs an SNR of 85 dB and a NF of 17.8 dB. The US 1900 MHz and European 1800 MHz bands only require an SNR of 75 dB and a NF of 27.8 dB. There are many ways to implement a receiver design, so these numbers may vary depending on the NF and conversion gain of the analog front end.

While software-defined radio may not be the optimal solution for every cellular market, there are clear benefits of this technology in existing markets. In the US 1900 MHz frequency band, products such as the AD6654 are available today to meet the technical requirements, eliminate components, and reduce cost. As new high-performance products are developed, operators will be able to further reduce the cost of migrating from GSM/EDGE to UMTS. The European market can realize these same cost benefits, but technical limitations of the 900 MHz band will continue to drive the trend of improving SNR and SFDR in analog-to-digital converters.

Chris Cloninger is Systems Applications Engineer, Wireless Infrastructure Products at Analog Devices Inc. (Norwood, MA)

See related chart