Testing E-OTD

An emulation-based approach can test 3GPP Enhanced Observable Time Difference capability in GSM/GPRS and W-CDMA handsets.

By Darcy Smith, Agilent Technologies -- Test & Measurement World, 12/1/2008 2:00:00 AM

To ensure that the 3GPP’s E-OTD (Enhanced Observable Time Difference) capability found in today’s GSM/GPRS and W-CDMA mobile handsets operates as anticipated, you need to test the handset’s performance. The simplest way to do that is to emulate only two cell towers with a known, but fixed, offset in timing between them. This approach could serve as a quick check, but it falls far short of a real characterization.

For a detailed characterization of E-OTD operation, manufacturers must emulate multiple cell towers, mobility, and variable channel conditions (see “Location-relevant information and E-OTD”). One plausible means of accomplishing a thorough characterization is through the use of a test system that emulates 12 cell towers with independent time-delay elements and variable channel conditions. As part of the test setup, a channel simulator, often called a fader, provides the time delay from the cell to a phone as well as the fading channel conditions. Just as a real network must calibrate the actual system time, the test system must calibrate the timing of each cell emulator, including both the timing from frame-to-frame and the offset in frame numbers from cell-to-cell.

Using such a test system, you can easily perform tests for a stationary handset. You can hypothetically place the cell towers at known locations on an x-y grid, representing longitude and latitude, and you can place the handset, either deterministically or randomly, at some spot on the grid. The test system computes the propagation delay from each cell tower to the handset location and enters it into the delay elements of the channel simulator from each cell. In this way, the arrival times at the handset are delayed as if in a real system. The handset makes its E-OTD measurements and computes its location by using the data from the handset and the system calibration values. You can then verify the results against the location of the handset on the grid. Alternatively, in assisted mode, the mobile handset reports the timing measurements back to a network entity called the SMLC (Serving Mobile Location Center), which in turn uses them to determine the handset’s location.

To create variations in this stationary test scenario, you can change the settings for variables such as the amplitude from each cell, the channel fading from each cell, and the interference from distant cells (on any of the measurement channels).

For a typical nonstationary test scenario, you can emulate the slow movement of the handset in the network by changing the time delay and signal levels from the different cell towers. As cells fall behind the coverage path of the handset, they are reassigned ahead of the handset’s path with new x and y coordinates and are slowly brought back into the handset’s range.

Figure 1.  In this test system configuration, an 8960 wireless communication test set serves as the emulator for individual base stations; a channel simulator mimics channel delay and fading. The test set can be used as the serving cell in an E-OTD test system to issue location requests to a mobile handset and then receive location estimates or measurement results in response.

In either the stationary or nonstationary scenario, the control interface between the cell tower and handset for E-OTD control and measurements is generally carried on a message similar to an SMS (short message service). The cell uses a downlink message to command the handset to make an E-OTD measurement. The results are then reported on the uplink using these special SMS messages.

To accommodate various use scenarios and requirements, you can configure a cellular system simulator by using 12 base-station emulators (Figure 1) configured for either GSM/GPRS or W-CDMA operation. You can simulate changes in the position of the mobile handset by varying the delay from each cell tower to the handset. This delay is controlled by the time delay in a channel simulator.

Calibrating the W-CDMA test system

Figure 2.  The top diagram depicts the relative time of three hypothetical cells, while the bottom diagram shows the corresponding hypothetical frame timing.

Testing the E-OTD capability in a W-CDMA handset can be especially challenging, due to the difficulty in finding the relative frame numbers and frame-time offset. To better understand how to deal with this challenge, consider a transmission from three hypothetical cells and the BCH (broadcast channel) framing. The BCH—a logical channel that is 20 ms in length and carried on two consecutive physical frames—contains the frame count parameter used in E-OTD along with the time of arrival of the frame start.

In Figure 2, the top diagram shows the relative timing near frame 0 from cell 1. Here, you must determine the values for x and y (integer frame offsets) and the delta between the start of the frame between cell 0 and the other two cells. This delta time must have precision in the tens of nanoseconds.

In W-CDMA, the frames are counted from 0 to 4095 (12 bits). Because the BCH requires two frames to transmit a signal, the value of the frame counter increments by two and contains only even values. In fact, the LSB (least significant bit) is not transmitted, only the upper 11 bits. The real frame count, at the starting frame of the BCH, is calculated by multiplying 2 by this 11-bit number.

Also in Figure 2, the bottom diagram depicts a hypothetical frame timing of the same three cells. The graphic shows the offsets in the 0 frame for the three cells.

In this example, the accurate calibration requires:

• 50-ns (maximum) uncertainty per cell tower. This assumes the worst-case addition of errors to yield 100-ns uncertainty in a single difference measurement.

• 20-ns per cell frame-to-frame timing. The offset of the frame number must be exact; an error in this value equates to a 10-ms offset, which is unacceptable.

Real-world example

You can develop an E-OTD test setup with commercially available instruments, such as the Agilent 8960 wireless communications test set. Using the 8960, you can generate a trigger signal that is synchronized to the frame counter and is at a constant offset from the frame clock. If you use separate test sets to emulate the serving cell and a neighbor cell, you can measure the time interval between the trigger of the two cells to determine the integer and the fractional delay between them.

The difference in frame numbers is given by integer multiples of 10 ms in offset, which represents the difference in frame-to-frame timing, assuming constant timing between the frame rollover trigger signal and transitions in the RF modulation. Generally speaking, the time-delay measurement has these requirements:

• The input must consist of 12 separate TTL signals for a system emulating 12 towers.

• One of the TTL signals (always the same one) must be used as a start timer for all measurements. Each of the other 11 can be considered a stop for its own timer.

• W-CDMA triggers must consist of a single pulse occurring approximately every 42 s. If the 8960 test sets have been turned on at random times, the timing can be considered random across all 12 units. Consequently, there can be as much as a 42-s delay after the system is armed before the start trigger occurs and up to another 42-s delay before the last unit sends its stop trigger.

• You must configure the trigger out of unit 1 to start a time-interval measurement, and you must configure the triggers out of the other units to be stop interval triggers.

• The test-time range needs to go up to 42 s, with a resolution and accuracy of approximately 1 ns, requiring 11 digits and a long measurement period.

The test process outlined in “E-OTD test steps” speeds up the test process, forcing all measurements to occur within a 10-s window (the entire test takes about 50 s) and dropping the timing resolution requirements to 10 digits. As a tradeoff, however, all frame count numbers must be in close alignment.

The timing measurement for GSM handsets is similar to that for the W-CDMA process, but the GPIB command for GSM sets the counter approximately 1 s before its frame rollover. This eliminates the delay in step 5 of the test process. Also, it establishes the initiator as a dedicated frame-reset command on GPIB to test test set 1 in step 1 and to test the rest of the test sets in step 3.

With the market for location-based services continuing to grow, technologies like the 3GPP’s E-OTD have become increasingly important. Being able to adequately test this position-location capability in GSM/GPRS and W-CDMA handsets is equally important and requires a test system capable of making accurate radio timing measurements. The test system I have proposed offers a plausible means of accomplishing that goal and, as a result, can play a key role in ensuring that E-OTD-enabled GSM/GPRS and W-CDMA handsets perform as expected.