Testing and certification of Cognitive Radio equipment

After reading a recent discussion on the Linked-in Cognitive Radio group I researched on the testing procedure of Cognitive Radio equipment. It is not straightforward to define a testing procedure so that reconfigurable systems are guaranteed to comply with regulations under all circumstances: the behavior of a node may depend on the current conditions and past history of the environment.

Right now there is no standard procedure that allows vendors to test their cognitive radio prototypes. In this sense the Wireless Innovation Forum's Test and Measurement Work Group is currently preparing a report called "Test Guidelines and Requirements for Television Band Devices (TVBDs) Designed to Operate on Available Channels in the Broadcast Television Frequency Bands":

The report will identify unique test challenges created by SDR/CR radio system technology used for TVBDs and will provide a basis for test and certification."

These challenges include dynamic waveform activation, opportunistic scheduling, policy based operation, spectrum sensing and interference avoidance. While this report will be an useful tool, its final version is scheduled in October 2010.

Other tests have been performed so far they have been oriented to measure only a subset of parameters of the system or to show the proof-of-concept of the unlicensed access paradigms. For example, the FCC testing of TV white space devices developed a procedure for the evaluation of equipment of different vendors, however it was oriented to the elaboration of the rule-set for unlicensed use of the tv-band. The test plan divides the procedure in three well differentiated parts:

1. Laboratory bench tests of sensing capabilities (single laboratory-grade DTV, single RF-capture DTV Input Signal, multiple DTV Signals and wireless microphone signals).
2. Laboratory bench tests of generated interference (transmitter emissions characterization and interference to wireless microphones).
3. Field Tests (both for sensing performance and generated interference).

The results of this testing plan conducted to the geolocated database based approach finally approved by the FCC.

The DARPA XG report "A Description of the August 2006 XG
Demonstrations at Fort A.P. Hill" presents a detailed description of the tests carried out in an outdoor six node network of Next Generation (XG) radios capable of using spectrum over a wide range of frequencies on a secondary basis. The conclusions read as follows:

A six-node XG network and three two-node XG networks were tested using four test scenarios: two were without XG adaptation and two were after adaptation. The audience, members of the U.S. government's military and spectrum management officials, was able to see clearly how XG could adapt to the local interference landscape and share frequencies rapidly and with little to no interference with existing legacy military radio systems. This demonstration showed that DARPA concepts of frequency sharing in a mobile environment are possible with today's technology."

On the other hand, Europe is also defining the requirements for the unlicensed access to TV-band and considering the possible regulatory issues in the SE43 Regulatory Affairs Working Group:


While this study is not mature enough to offer a practical testing and certification protocol, Ofcom (communications regulator in the UK) presented in February 2009 a consultation regarding the exploitation of cognitive radio in digital dividend bands:

Determining the sensitivity of a device is relatively straightforward. It is often quoted in device specifications and can readily be confirmed in laboratory trials. For example, a number of DTT receivers could be procured, a test signal inserted into their aerial socket and the strength of this signal reduced until the picture quality visibly deteriorates."

However, the report also stress the fact that the sensing performance has to be achieved in real world scenarios:

In particular, the devices must be able to sense in the presence of strong signals in adjacent channels (or they must choose not to use channels where strong signals are present in adjacent channels). They must also be able to sense adequately in the presence of other cognitive devices using nearby channels. Hence, we suggest that, as part of the specification and resulting type-approval verification, cognitive devices be tested to ensure that they do not incorrectly declare a channel to be unused regardless of the level of signal in adjacent channels."

Other points commented in the document are Bandwidth, Signal characteristics, Time between checking for channel usage, Out-of-band performance, Politeness...

As a conclusion, while there exists no integral test plan for Cognitive Radio devices, the evaluation of some of their capabilities can be carried out following a test methodology similar to the ones presented. For example, sensing and interference can be determined for single nodes using a procedure similar to the FCC test plan. Other more sophisticated network features require more complex tests that measure the aggregated interference and the dynamic capabilities of the network, such as the DARPA XG evaluation or some of the points considered in Ofcom's report.