Part 1 of this two-part series focuses on the software-defined radio architecture, key applications, and analog-to-digital conversion. In part two, which will appear next week, we will explore digital-to-analog conversion, receive signal processing, and transmit signal processing in software-defined radio designs.
By: Brad Brannon, Senior Systems Application Engineer, Analog Devices
Uses for software radios
Wireless standards evolve on a daily basis. Standards bodies, such as the Telecommunications Industry Association (TIA) and European Telecommunication Standards Institute (ETSI), are continually reworking current specifications in order to provide protocol improvements, capacity gains, and performance enhancements. These continual changes, however, can cause costly headaches for today's network operators. In order to keep pace with new standards, operators must continually upgrade their infrastructure products to support each and every protocol change, which can be a costly venture.
In order to help operators ease the standards update process and curb cost, many base station designers are trying to perform more functions in software as opposed to hardware. One emerging area they are focusing their attention is in the radio portion. Thus, the need for software-defined radios is growing in base station designs. Let's start by examining the uses of software radios
Uses for software radios
Software-defined radio systems have many potential uses. Whether the application is land mobile radio (LMR), military, or cellular, the requirements are quite similar. A software-defined radio is one that cannot only change its frequency, but bandwidth and modulation scheme through software programming. Of course, any radio could be software programmable through selection of a variety hardware elements. But these solutions would only provided a limited range of flexibility. True software programmability offers an unlimited range of operations.
To offer an unlimited range of operations, several things must happen. First, data must be manipulated with a high MIPS digital signal processor (DSP), which forms the heart of a software radio. In addition to a general-purpose DSP, numeric accelerators are required to process the carriers digitally. These accelerators provide extremely high MIPS that perform the fixed functions of RF carrier modulation, tuning, and filtering, all in the digital domain. The final interfacial elements are the data converters.
In a true software-defined radio, the receiver front end consists of an analog RF downconverter that converts the desired signal band to a convenient IF frequency for digitization (Figure 1). The downconvert is followed by a high performance analog-to-digital (A/D) converter, which digitizes the IF spectrum. The A/D converter output is processed with a receive signal processor (RSP) which is responsible for tuning and channel filtering. The output of the RSP consists of a channel filtered digital IF signal requiring only demodulation, which the DSP handily provides.
In the software radio's transmit path, the DSP sends modulating digital data to the transmit signal processor (TSP). The TSP then modulates the digital carrier. Data is then converted to the analog domain using a high performance digital-to-analog (D/A) converter, and mixed up to RF.
Software radios are most suitable for multi-carrier applications where redundant hardware exists. Consider the example of a triple-mode CDMA receiver capable of receiving one IS-95 channel, three adjacent IS-95 channels, or one 4.096 Mchips/s channel.
The traditional implementation for a triple-mode base station transceiver system (BTS) would require that each mode have a separate signal-processing path. Thus a path would need to exist for a single IS-95 carrier, one for three adjacent IS-95 carriers, and one for a single 4.096 Mchip/s carrier. Each signal path would also need a separate IF filter and amplification stage. Switching between modes would require that the correct signal path be selected. If implemented using software radio techniques, the three analog filter paths would be replaced by re-configurable digital filters capable of meeting the various needs of these three standards.
Software-defined radios can be used with any type of modulation including any current analog or digital wireless modulation standards. Furthermore, since the core of a software radio is its DSP, and assuming that the RF hardware in front of the DSP is properly designed, the same receiver can be used for both analog and digitally modulated signals, simultaneously if necessary.
The key is the software used to run the DSP itself. As data converter and DSP technology continue to evolve, radio technology will improve as well.
The fact that a radio is software programmable offers many benefits. First, from a manufacturing point of view, radio manufacturers can design a generic radio in hardware. As air interface standards change (as from FM to CDMA or TDMA), manufacturers are able to make timely design changes to their radio simply by reprogramming the DSP.
From a user or service providers point of view, the software radio can be upgraded simply by loading the new software at little or no cost, while retaining all of the initial investment. Additionally, the receiver can be tailored for custom applications at very low cost, since only software costs are involved. Also, since these transceivers can easily be paralleled, they make ideal candidates for phased array technology now being developed.
In order to achieve these goals, several key components must meet certain performance goals. Since a software radio requires more digital processing than a traditional radio, the signals must remain digitized longer throughout the system. Therefore, high quality data converters are vital to this architecture. Likewise, digital processors are required to process data to and from the data converters.
There are two key specifications for an A/D converter for wireless applications. The first specification is signal-to-noise ratio (SNR). SNR is the ratio of signal energy to noise energy and is very useful in computing sensitivity of a receiver. What noise figure (NF) is to an amplifier, SNR is to an A/D converter. The noise in SNR consists of thermal noise and quantization noise (the noise generated when a signal is digitized). In and ideal A/D converter without thermal noise, the SNR is equal to:
SNR = 6.02N + 1.8
However, many things can effect how an A/D converter performance. Already mentioned is thermal noise. Other effects are non-ideal quantization (imperfect A/D converter) and aperture uncertainty (a wide band phase noise on the clock) in the sample clock. If all of these effects are considered then the simple equation for SNR of the converter becomes:
In most receiver designs, the designer seeks to use NF. Although A/D converters are voltage devices and NF is a power measurement, an equivalent NF can be calculated based on a full-scale referenced SNR. For example, with a full-scale SNR of 75 dB and a full-scale of +4 dBm, the integrated noise within the ADC is -71 dBm.
When measured over 1 Hz, the noise power per Hz is reduced by 10*log(Fencode). Therefore, if the clock rate is 80 MSamples/s, then the noise is spread over 40 MHz (the Nyquist bandwidth) which gives a noise per Hz of -147 dBm. The theoretical noise in a 1 Hz band is -174 dBm. Therefore, the noise figure for this example is 27 dB (-147 dBm - -174 dBm)
The second specification of interest for an A/D converter is spurious response, which consists of 2nd, 3rd, and higher order harmonics. Spurious performance is very closely related to the air interface attempting to be achieved. For example, if the multi-carrier receiver is attempting to receive a GSM signal, the air interface requires a 91 dB signal dynamic range plus headroom.
This specification indicates that one in-band carrier may exist at the A/D converter full-scale, while another exists 91 dB smaller. If the harmonic performance of the A/D converter fails to meet the air interface requirements, then the harmonics of the larger signals will interfere with the weaker signals.
About the author:
Brad Brannon, senior systems application engineer, Wireless Infrastructure Products, Analog Devices, 7910 Triad Center Dr., Greensboro, NC 27409. Tel: 336-605-4212; Fax: 336-605-4332