As a result of this requirement for application-specific measurement solutions, original equipment manufacturers (OEMs) are beginning standardize on modular "software-defined" platforms for test and measurement products. These platforms enable technicians to re-program the measurement system either at the factory or dynamically in the field. This way the devices can support several different air interface standards and configurations based on the requirements of the communications device.

The mobile wireless communications market has three parts: peer-to-peer communications devices, multi-hop wireless network devices, and single hop wireless network devices.

First, peer-to-Peer communications devices represent the simplest of the three categories. These devices include broadcast and point-to-point radios. Broadcast devices enable any radio tuned to a specific channel to communicate with any other radio that is within range and tuned to the same channel.

Point-to-point radios are digital communications devices that incorporate some type of security feature, such as data encryption or frequency hopping, to ensure private communications between two specific radios. Peer-to-peer communications devices typically incorporate one or more of the standard modulations formats, such as AM, FM, FSK, PSK, and QAM, at various data rates. When more than one transmission scheme is available, the modulation format and data rate for a particular communications session are either selected by the user or dynamically set via a quality of service measurement.

Testing for peer-to-peer communications devices represent the baseline testing for all communications systems. These tests are limited to one channel at a time and include transmission quality, bit error rate performance, and power level control. The challenge for test sets targeting these types of radios is to ensure sufficient flexibility to validate each of the modulation formats and data rates supported by a particular device.

Second, multi-hop wireless network devices also establish peer-to-peer communication links among radios. The difference involves this communications scheme; each radio supports more than one peer-to-peer link and acts as a repeater or bridge between the established links. This results in a dynamic, infrastuctureless network supporting potentially hundreds of mobile units. An example of this type of ad-hoc network is the military tactical packet radio network. Specific radios within this network may also be configured to act as network gateways by supporting more than one waveform simultaneously. This allows seamless communications between mobile units on disparate networks.

Testing of a multi-hop network device is significantly more complicated than the test of a peer-to-peer communications device. In addition to the baseline communications tests, test sets for multi-hop network devices must also validate features specific to the operation of the device within the network, including link establishment and packet routing. For gateway nodes, each of the supported waveforms must be tested, and the movement of data between the two waveform networks must be verified. Finally, testing may be needed to assess the overall performance of the device within the network. This includes:

• network reconfiguration time;
• number of nodes that can be supported;
• average throughput/latency; and
• average number of packets lost.

Those testing these parameters must tailor them to the specific network architecture, and require significant flexibility and scalability on the part of the measurement system.

Finally, a single-hop network uses a fixed infrastructure consisting of one or more centralized wireless gateways or base stations. Mobile units in this type of network communicate exclusively with the base station, which routes traffic between the mobile station and the rest of the network. Single hop networks include commercial cellular, wireless local loop, satellite systems, and tactical airborne communication nodes.

Separate test sets are typically required for mobile stations and base stations in a single hop network, with testing that includes the baseline communications tests presented above and network centric tests such as:

• mobile/network registration and synchronization;
• call setup and tear down;
• handover of a mobile from one base station to another; and
• coexistence of the single hop wireless network device with other networks in the same location.

Complicating test of these parameters is the mobile unit's need to support communications on several different networks. For example, a cellular handset may support the IS-136 TDMA standard while in North America with roaming support for both AMPS and GSM. The mobile unit test set must therefore provide simultaneous evaluation of each of the air interfaces specified.

Software Defined Radio
An examination of the challenges presented above indicates several features that are common for measurement solutions targeted at communications devices in each of the perspective markets. These include flexibility, modularity, scalability, extensibility, and evolvability.

First, the test set must be capable of supporting several different waveforms/air interface standards simultaneously. Then it must consist of modular hardware and software that can be reconfigured to meet the specific measurement requirements of the communications device. Next, the test set must scale to accommodate both limited testing of the communications link and comprehensive testing of the device in a fully loaded network environment. After that, it must integrate easily with other types of test equipment to provide extended measurement capabilities. Finally, the test set must incorporate easily upgradeable components to allow the end user to protect the investment in the test and measurement equipment.

The software defined radio addresses each of these requirements for wireless test and measurement. These devices "provide software control of a variety of modulation techniques, wide and narrowband operation, communications security functions (such as hopping), and waveform requirements of current and evolving standards over a broad frequency range," according to the Software Defined Radio Forum.

Historically, engineers would implement a generic wireless test and measurement system in a "stovepipe" design with one air interface standard in mind. It would use any programmable device in the system evaluated primarily on a cost basis.

Yet a modular software defined radio measurement platform would enable uses to reconfigure each of the major functions of the test set on the fly, as well as support several different air interface standards and several different test configurations.

The core of the wireless test and measurement system is the digital transceiver subsystem. On the receive side, this subsystem receives the digitized RF bands, extracts the channels of interest through the use of software defined digital down converter, and then demodulates these signals on a digital signal processor or general-purpose processor to extract the payload data for follow on processing.

This process reverses on the transmit side as it modulates payload data on the DSP or general purpose processor and then up-converts in the channelizer for retransmission. Test experts can evaluate the performance parameters for a digital transceiver in a generic wireless test and measurement system through case studies of several different measurement solutions.

Engineers at Spectrum Signal Processing in Burnaby, British Columbia, have developed a modular software defined digital transceiver architecture that consists of three modules. The I/O transition module interfaces directly to the RF subsystem and acts as the analog/digital interface for the system. This module supports as many as 800 mega-samples per second (MSPS) half duplex or 600 MSPS full duplex communications with the software defined I/O processor.

The software defined I/O processor (SDIO) consists of five Virtex II processing elements and is available in several different configurations. This module provides the channelization and preprocessing functions of the digital transceiver system with support for as many as 280 channels of 2 MHz bandwidth or less. The SDIO module can communicate with as many as five baseband processing engines at a composite rate faster than 400 megabytes per second. Two baseband processing engines provide modem processing; one incorporates as many as six TMS320C6416 DSPs and one incorporates two to four MPC7410 general-purpose processors.

Key to the software defined digital transceiver architecture is its passive backplane architecture, which uses three independent buses based on the dataflow requirements in each section of the transceiver subsystem. The I/O bus uses a fast serial protocol, such as Serial RapidIO, to pass data as fast as 2.8 gigabytes per second from the transition module to the SDIO module.

The fast bus uses a passive backplane architecture to move data between the SDIO modules and the baseband processing engines. This bus also uses a serial protocol and moves data faster than 400 megabytes per second. Finally, the payload data bus uses a 100-mebabit-per-second Ethernet in an embedded CompactPCI packet switch backplane, as specified in the PICMG 2.16 draft standard, to allow communications between the digital transceiver and the rest of the test and measurement system.

To illustrate, consider an airborne communications gateway configured as a communications backbone to provide beyond-line-of-site connectivity between several different tactical networks and tactical operations centers.

This system allows several different users to connect with the gateway at random, with the gateway load never greater than 75 users at a time. Support for as many as 17 different waveforms is necessary for this gateway, with data rates distributed as 40 voice and low-speed data users with data rates as fast as 64 kilobits per second; 30 multimedia data nodes with data rates as fast as 512 kilobits per second; and five high capacity backbone nodes with data rates as fast as 8,192 kilobits per second.

Testing this system must include an automated study of each of the possible waveform configurations at various network loads.

To identify the channelizer to modem bandwidth requirements for this system, experts must choose worst-case waveforms for each of the data rate categories. For voice and low-speed data assume binary phase shift keying (BPSK) modulation and rate 1/2 convolutional coding for forward error correction. This means that the data rate is 64 kilobits per second x 2 (error correction) x 1 symbol/bit (BPSK) x 4 samples/symbol x 4 bytes per samples (16 bits per sample, I and Q) = 2.048 megabits per second for each user.

Similarly, for multimedia data, assume quadrature phase shift keying (QPSK) modulation with a rate 1/2 convolutional code. The channelizer to modem data rate for this waveform is 512 kilobits per second x 2 (error correction) x 1symbol/2bits (QPSK) x 4 samples/symbol x 4 bytes per samples (16 bits per sample, I and Q) = 8.192 megabits per second for each user.

Finally, consider the high-capacity backbone network, assuming 16-QAM modulation for this waveform with the rate 1/2 convolutional code. This provides an data rate of 8,192 kilobits per second x 2 (error correction) x 1symbols/4bits (16-QAM) x 4 samples/symbol x 4 bytes per samples (16 bits per sample, I and Q) = 65.536 megabits per second for each user.

So the total bandwidth required from the channelizer to the modem is: (40 x 2.048) + (30 x 8.192) + (5 *65536) = 655.36 megabits per second.

For processing, assume that modulation and error correction processing are all that is required on the baseband-processing engine. In addition, assume that it takes at most 100 operations per sample are required to demodulate and decode these signals, with a smaller number of operations per sample required to encode and modulate the reverse channel. As such, a composite estimate of 200 operations per sample is reasonable for each user. The total processing power required for each of the data rate categories are:

• 102.4 mega-operations per second per user for voice and low speed data;
• 409.6 mega-operations per second per user for multimedia data; and
• .3 giga-operations per second per user for backbone communications.

For this case study the quad MPC7410 baseband processing engine will be used. At 500 MHz, each of the MPC7410 processors can average over 3 GFLOPS providing a composite processing power of 12 GFLOPS per board. If this processing power estimate is cut by a third to allow for book keeping and memory swaps, this means that each baseband processing engine can host around 87 voice and low speed data modems, 20 multimedia modems, or 2 backbone modems.

One possible configuration for this test and measurement system is shown in figure 6 below. This system consists of three clusters, one for each data rate category. The maximum data rate out of the SDIO module to the baseband processing engines is in the high capacity backbone cluster, with a sustained rate of 327.68 megabits per second required.

Lee Pucker is the systems architect at the Spectrum Signal Processing Inc. Wireless Systems Business Unit in Burnaby, British Columbia.

by Lee Pucker
June, 2001

http://mae.pennnet.com