Software radio technology solves the compatibility problem of wireless communication

tmily

Software radio technology solves the compatibility problem of wireless communication [Copy link]

Portable wireless devices may be converging, but their interfaces continue to separate, or "evolve," if you will. For a phone to survive a trip from the U.S. to Europe and Asia, it needs to be compatible with an ever-growing variety of different protocols, modulation techniques, and frequency bands, and to be able to recognize them and seamlessly adjust and match them. For handset manufacturers, that's a challenge; for base station vendors, it's a nightmare.
How difficult is it? "A few years ago, dual-band GSM phones satisfied most of the market," says Bill Krenik, senior director of wireless research at TI. "Now the chipsets are at least GSM/GPRS, and soon they will most likely be GSM/GPRS/EDGE, covering PCS and DCS; join UMTS, and add WCDMA support for the IMT-2000 band at 2,100MHz. On the modulation front end, you need to handle GMSK for GSM, higher-order modulation of 8-PSK for EDGE, and QPSK for wideband CDMA; the next challenge is HSDPA, HSUPA, and 16-QAM; then OFDM-based The LTE (Long Term Evolution, or 4G) standard is targeted for the 2010 time frame. By then, we will need to support TDMA, CDMA, and OFDM; support at least six different frequency bands, and most likely more; and support modulation techniques from GMSK to QPSK, 16QAM, and 64QAM. We need radios and base stations that can address all of the above technical issues." At this point, software-defined radio (SDR) technology is more attractive than many hardware-based RF signal chains.

Enter SDR
The concept behind SDR is simple: start with a high bandwidth, high linearity FR front end, then direct its output to a high speed ADC, and then use a high speed DSP or FPGA to do all the signal processing in software. The word "high speed" mentioned many times in the previous description is exactly where the challenge lies.
A software defined radio is a wideband transceiver whose characteristics are defined by software. The ITU defines SDR as: "A radio system that can set or modify RF operating parameters, including but not limited to frequency range, modulation type or output power, through software and/or technology capable of performing the same function." SDR can minimize the hardware required for mobile phones or base stations, while also reducing the need for "future proofing" infrastructure corresponding to evolving wireless standards. SDR basically turns a receiver into a computer with an RF front end. However, SDR requires a very fast computer to oversample the various multi-GHz signals and process all of them in near real time.
SDR is not an extreme method without intermediate processes. The SDR Forum defines a progression from "hardware radio" - the current approach - to "ultimate software radio", with a number of intermediate technology stages in between:
● Level 0 - Hardware radio. No modifications can be made to the system. System operation is accomplished by switches, dials and buttons.
● Level 1 - Software Controlled Radio. Control functions may be implemented in software, but these wireless communication devices cannot change characteristic parameters such as frequency band or modulation without changing the hardware.
● Level 2 - Software Controlled Radio. This radio technology is controlled by software and can provide a wide operating range without any hardware modifications. It generally includes a separate antenna before broadband filtering, an amplifier and a downconverter before A/D conversion. When transmitting, the baseband signal enters the DAC and is converted to an intermediate frequency (IF) frequency, which is then heterodyned to the desired output frequency before filtering and entering the power amplifier. Although the bandwidth of the front end is a limiting factor, a fairly wide frequency range can be controlled using software because SDR technology can provide multiple demodulation techniques in both wideband and narrowband operation. SDR technology can store a large number of waveforms or air interfaces and add new content by uploading from disk or downloading from space.
● Level 3 - Ideal SDR. This SDR connects the output of the RF front end directly to the ADC and then to the DSP, eliminating most of the analog components, thereby reducing distortion and noise.
● Level 4 - Ultimate SDR. This SDR has no external antenna and no operating frequency or bandwidth restrictions. It directly transmits the digital baseband output signal and detects and converts between air interfaces in a few milliseconds. This is a low-power and extremely fast SDR, but it is unlikely to appear in the form of a commercial product before the child who is learning to speak graduates from college.

From antenna to DSP
A typical SDR system (as shown in Figure 1) consists of various RF front ends that provide input signals to a wideband ADC. From the ADC, the signal enters the programmable DSP engine. In transmit mode, the digital baseband signal first passes through the DAC and intermediate frequency/radio frequency (IF/RF) upconverter before being passed to the input power amplifier (PA) and antenna.

Figure 1. A typical SDR receiver.

Building a configurable, wideband, highly sensitive, and even bulletproof RF front end is where things get complicated. “Even if you’re looking at small signals, you have to provide a lot of data processing power,” said Doug Grant, director of business development for RF and wireless systems at Analog Devices. “When you get signals of different bandwidths, you have to see both the widest and narrowest bandwidths and accommodate enough dynamic range. It all comes down to power, because to get a wide dynamic range for the wireless front end, you have to increase power consumption.” If the front end is corrupted by a large out-of-band signal, no amount of downstream filtering can correct it. “Power is consistently cited as the biggest obstacle facing any approach to SDR,” said Grant.
In an ideal SDR system, the output of the RF front end would go directly to the ADC before processing. However, the ADC is one of the main challenges facing SDR. At 1GHz, the upper limit of the ADC’s dynamic range is 20 bits or 120dB. At higher frequencies, the problem becomes more challenging because the ADC’s sampling rate must be at least twice the highest frequency (fmax) to meet the Nyquist sampling theorem. Therefore, the receiver generally passes the signal through a bandpass filter to remove unwanted signals, and then uses heterodyning to reduce the signal frequency to a range that the ADC can handle. The signal is then passed through a low noise amplifier (LNA) and into the ADC. Super heterodyne receivers with multiple RF front ends and antennas are likely to appear in the future.

Figure 2. Software-controlled tasks for SDR.

Pure digital RF architecture, such as TI's digital radio processor (DRP), has an ADC on the antenna, and the DSP does all the rest. Since the analog part is eliminated from the RF circuit, this architecture requires a high-bandwidth Σ-Δ ADC and a very fast DSP, both of which consume a lot of power, so this approach is more suitable for base stations, not for mobile phones.

Software Processing
Once the signal enters the digital domain, the DSP engine can perform a considerable amount of processing, including handling different modulation types, channel access, spread spectrum processing, network interface definition, security, beamforming, forward error correction and digital down/up conversion.

Figure 3. Software architecture of SCA.

The software architecture of SDR is defined by the Software Communication Architecture (SCA; see Figure 3), which was originally developed in the 1990s for the U.S. military's Joint Tactical Radio System (JTRS) to allow all military forces to communicate with each other in real time to get information about fires, police, ambulances, and forestry personnel in each other's area. (Believe it or not, they didn't do it.) SCA is an open architectural framework that tells designers how hardware and software elements work together in the JTRS SDR. The SCA specification specifies the software operating environment that can configure waveforms and details the interfaces that waveforms must support. It also details the operating system OS (POSIX), middleware (CORBA ORB), and interface framework. The JTRS group and the SDR Forum are working with the Object Management Group's SWRADIO Special Interest Group to establish an open international industry standard for SDR systems based on SCA. In the meantime, SCA has become the de facto international standard for SDR.
JTRS SCA 3.0 adds a hardware abstraction layer (HAL) to the signal processing subsystem (SPS). SPS provides high-speed computation for OSI layer 1 (modem, extension, code) and application-level functional blocks (audio, video) of SDR. SPS can be implemented using a combination of DSP and/or FPGA.

The Need for SDR Signal Processing

Xilinx and Altera both argue that "digital signal processing" is not necessarily synonymous with DSP. The signal processing required for high-frequency commercial SDR receivers is in fact daunting (see table). Manuel Uhm, senior DSP marketing manager at Xilinx, argues that it makes sense to use MCUs in control blocks, DSPs in low-MIPS applications, and FPGAs in high-MIPS applications. Xilinx has made a JTRS SDR tool that combines a CORBA ORB with an RTOS for information passing, an application layer for narrowband and wideband waveforms in an SCA core framework, and a physical layer (EMAC, PHY, digital up- or down-converter) in an "SCA-enabled SoC" (aka Virtex-4).

Current Applications of SDR
We are still far from the ideal SDR, but commercial SDRs have begun to be used in multimode base stations, and these products use at least some SDR technology. Grant of Analog Devices said: "We have made some progress, but we have to limit the definition of SDR. If it means that programmable signal processing completes all demodulation, equalization and detection processes except the air interface, then we have done it. If you add 'using a general radio front end' to the definition, then we are not done yet."
The first company to market an FCC-approved SDR base station was Vanu Inc. It established the Vanu Software Wireless GSM Base Station in 1994. Vanu took a different approach to implementing SDR: running portable application software on a computer using a general-purpose CPU, and the signal processing is completely done by software. John Chapin, Chief Technology Officer (CTO) of Vanu, said, "You need to build the system you envision: the real value added is the software. We wrote portable code that can be ported to the next generation of processors without much cost." Chapin pointed out that using Intel processors is simpler than using DSPs or FPGAs. Vanu's next-generation Anywave GSM base station will run on Intel's Xeon dual-core processor platform.
Mid-Tex Cellular is the first US cellular operator to operate based on SDR. Mid-Tex built Vanu's software radio base station on HP's ProLiant platform running Linux, which enabled it to add GSM and WCDMA capabilities to its rural 800MHz TDMA system. The software base station can run multiple spatial standards simultaneously, dynamically allocating the system to support both 2G and the developing 3G standards.
In addition, SDR has been applied to passive RFID systems. ThingMagic's Mercury4 is a multi-protocol RFID system that uses SDR technology to read any tag at the same time, such as EPC Class 1 and EPC Class 0 tags, rewritable Class 0 tags, ISO 18000-6B and UCODE EPC 1.19 tags. Mercury4 acts like a router, passing the tag information to the network, database and application software. This system is more like a base station than a tag reader.

Portable SDR?
How long will it take to move SDR from base stations to cell phones? Cell phones are a good market, but not the best destination. This is because of the power required for SDRs—both the RF front end and the GHz ADCs. Nevertheless, some multimode cell phones are emerging. Analog Devices has developed a dual-mode GSM/TD SCDMA device for the Chinese market. These devices have two different RF circuits, but all decoding and node detection algorithms are performed in software rather than hardware.
As the cost of computing power continues to decrease and the computing power of cell phones continues to increase, the day may not be far away when the SoC in a cell phone will include an integrated RF front end, high-speed ADC/DAC, high-speed DSP, non-volatile memory, and power management blocks. This is only possible with SDRs.