David A. Case NCE, NCT
On February 3rd 2006, after three years of work the US ITAC-R group addressing the development of the proposed US DFS test procedure to be used to certify products under FCC Part 15 was completed. This test procedure is milestone in verifying the sharing of the 5GHz band with Radiolocation systems.
The work group consisted of Industry experts as well as government experts from the Federal Communication Commission, National Telecommunication and Information Agency, Dept of Defense and other governmental agencies.
The Need for DFS
The requirements for DFS can be found in ITU-R Resolution 229, which addressed allocating the 5150 – 5350 MHz and 5470 -5725MHz for Mobile systems. This allocation on a non interference basis includes Wireless Access Systems including RLAN.
In order to facilitate band sharing, two cognitive radio techniques were adopted. The first is TPC (Transmitter Power Control), the second is Dynamic Frequency Selection.
As defined only the master (control system) needs to have DFS capabilities though client cards that support peer to peer must also be DFS capable.
Basically DFS functionality performs a channel availability check before using the channel and then listens in the quiet periods to determine if a radar signal is present. If one is present on a given channel the master which has DFS capabilities issues a control signal and moves the clients off the channel that has the radar signal to avoid interfering with the radar. As part of DFS, the systems also makes sure the channels are loaded evenly.
Systems operating in the 5250 – 5350 MHz and 5470 -5725MHz are required to employ DFS technology. There is a grace period for systems operating in the 5250 MHz band but not for the systems in the 5.4GHz band.
DFS Test Parameters
The final draft test procedure as sent to the FCC includes conducted test procedure as well as radiated measurement procedure. Further an alternative method was developed to simulate the more difficult hopping radar sequence.
Unlike the standard FCC radio testing of just the EUT, this test requires a system test with client card to demonstrate that not only the system can detect the radar but it sends out the correct information to the clients to have them move off channel as needed.
A custom software file to properly load the system at 50% is required to be used. This program can be downloaded from the NTIA site as well as other information needed to complete the test suite.
Unlike other Part 15 testing where we use the highest gain of each family, this test requires us to use the lowest gain of each antenna family to demonstrate compliance.
Radiated Test Set Up
The testing is basically transmitting data from the host to the client and then have it detect a number of different radar pulses to see if the system detects and moves off channel for the required time as well as perform the channel availability check while sending the data.
There are 3 basic radars that need to be tested, the short pulse fixed radars, the long pulse fixed radar, and the frequency hopping radar. Each type radar must be tested.
Verifying DFS Test Procedure
Once it was determined that the test procedure could verify that the DFS detection limits and operational characteristic worked, the next phase was to determine if DFS actually would detect radar.
In mid December 2005, NTIA tested several manufacturers systems detection capabilities against several US radars. This testing done at a military base demonstrated that the systems that were tested in the bench test and passed the DFS criteria could in fact detect the radar in questions. Further testing in Europe also proved this out.
Based on the results of both the bench testing and real world testing, the agreement on detection criteria was reached and the final results of this agreement are in the test procedure . The original dates for compliance were extended when the working committee could not reach closure on some issues.
FCC Approval Process
Since DFS is a new technology, the FCC will not allow a TCB to certify a radio with DFS functionality at this time. Therefore all radios being certified that are DFS capable will be required to go through the FCC review process.
The DFS procedure addresses specific information the report must contain as well as some specific formats for the data.
The review process will include a pre-grant audit to verify that the product DFS functions work. As part of the review process NTIA will have a review of the DFS test report and pre grant audit data and will have to sign off before the grant is issued.
This will extend the review process from the week TCB cycle to 60 or more days (I guess closer to 90). The FCC will also be doing post grant audits of systems once the products start shipping.
Until NTIA and specially FCC feel comfortable, the TCB’s will not be reviewing any systems with DFS.
There is no doubt in ones mind that the USG see this type of effort as way forward to address band sharing issues. The fact that the FCC and others are interested in Cognitive Radio techniques, it will take both side so the aisle those wanting to use the band and those who are required to share it to work together to develop test methods to insure sharing.
David A. Case NCE, NCT. He is a Senior Regulatory Engineer for Cisco Systems Corporate Compliance Group. He is a member of the US working group developing the DFS test procedure.
47 Code of Federal Regulations Part 15.407 (2005)
ITAC-R US DFS Test procedure Rev 7
ITU-R Resolution 229
Courtesy of Edgard Vangeel, EMEA Corp Compliance Cisco Systems. These are used in the US and ETSI Test Procedures.
US DFS Test Signals and Pass /Fail Criteria
Fixed Radar Set Pulse Width
(µsec) Pulses Minimum
Probability of Detection (Minimum Detections out of Minimum Samples) Minimum
1 – fixed 1 1428 18 60% (18 out of 30) 30
2 – variable 1-5 150-230 23-29 60% (18 out of 30) 30
3 – variable 6-10 200-500 16-18 60% (18 out of 30) 30
4 – variable 11-20 200-500 12-16 60% (18 out of 30) 30
Aggregate 80% (96 out of 120) 120
Long Pulse Radar Set Pulse Width
(µsec) Bursts Pulses Minimum probability of detection Minimum
1 50-100 1000-2000 8-20 1-3 [90%] [27 out of 30] 
Parameter Type Pulse Width
Length Pulses per Hop Hopping
Rate Pass/Fail Minimum Trials
1 – Fixed 1 333 300 ms 9 .333 kHz 70% (21 out of 30) 30