GSM RF Equipment Testing and Performance Analysis Ahmad H. Fares, Ali M. Khachan, and Ahmad M. Bakri Kasbah Department of Electrical and Computer Engineering American University of Beirut Beirut - Lebanon {ahf05, amk18, amb07}@aub.edu.lb Abstract: In this paper, we present four Radio Frequency (RF) measurements used to evaluate fundamental performance parameters of Global System for Mobile Communications (GSM) equipment. We describe the relevant theory for each measurement, and then we proceed to explain the algorithms associated with it. These algorithms are implemented using LabVIEW in a GSM Measurement Toolkit (GMT). GMT is developed for National Instruments (NI) to be used with PXI-5660 RF Signal Analyzer (RFSA) to test RF equipment. Using this toolkit, a Base Transceiver Station (BTS) downlink signal and a Mobile Station (MS) uplink signal are tested to ensure their conformance to 3rd Generation Partnership Project (3GPP) standards.
1. Introduction GSM is the most widely deployed mobile system with more than half a billion users spanning the globe. Measurements are indispensable for both GSM manufacturers and operators who are very concerned about the cost of test equipment. These measurements are used in quality control, calibration, and maintenance of both mobile and base stations [1] [2]. GSM is a trunked radio system in which the number of available channels is less than the number of possible users. This process of sharing channels among users is feasible because the probability of everyone demanding a channel at the same instance is very low. Multiple users of the system are granted access through the division of the system into frequency and time. GSM utilizes a combination of Frequency-Division Multiple Access (FDMA) and TimeDivision Multiple Access (TDMA), in addition to frequency hopping. The GSM frequency band is divided into 124 uplink/downlink carriers. Each carrier is divided in time into 8 time slots to allow at least 7 users to access the network using the same carrier. The process of testing consumes a lot of resources in terms of time and budget. NI’s virtual instrumentation technology promises a great reduction in test costs and enables the customer to administer the test setup and apply customized configurations. The four measurements, described throughout this work and included in the GMT, are adjacent channel power, modulation accuracy, mean transmitted RF carrier power, and transmitted RF carrier power versus time. For each measurement, we introduce its theory and illustrate its implementation in LabVIEW. The NI PXI5660 RFSA and GMT are used to test a BTS downlink signal and an MS uplink signal [3].
2. Adjacent Channel Power According to 3GPP standard, GSM adjacent channel power measurement is divided into two sub-measurements: spectrum due to modulation and wideband noise and spectrum due to switching. These two measurements are frequently referred to as output RF spectrum (ORFS) [5].
2.1. Spectrum due to Modulation and Wideband Noise Spectrum due to modulation and wideband noise measurement checks whether the modulation process is causing excessive spectral spread. In this test, one timeslot (except for time slot 0) shall be set up to transmit full power while all other time slots shall be turned off. In case of slow frequency hopping (SFH), any carrier may be used, else only the carrier transmitted shall be used [5]. First, the power is measured on the carrier frequency using a video filter with bandwidth of 30 KHz. The measurement shall be gated over 50 - 90 % of the useful part of a single time slot, and the measured value over this part of the burst shall be averaged. The averaging shall be over at least 200 time slots and only the active burst shall be included in the averaging process. The same above procedure is repeated at different offsets below and above the carrier frequency. These offsets are 100 KHz, 200 KHz, 250 KHz, and 400 KHz, in addition to offsets from 600 KHz to 1800 KHz in steps of 200 KHz. The test limits are expressed in relative terms (dBc), relative to the carrier power. As for the measurements performed on the different offsets, and according to 3GPP Technical Specification (TS) 05.05, the results should not exceed the limits shown in Table 1 for the BTS. The algorithm for this measurement is summarized in Figure 1. Table 1 BTS limits for spectrum due to modulation [5] Carrier Power level (dBm) ≥ 43 41 39 37 35 ≤ 33
Maximum relative level (dB) at specified carrier offsets (kHz) 100 200 250 400 600 to 1200 to < 1200 < 1800 +0,5 -30 -33 -60* -70 -73 +0,5 -30 -33 -60* -68 -71 +0,5 -30 -33 -60* -66 -69 +0,5 -30 -33 -60* -64 -67 +0,5 -30 -33 -60* -62 -65 +0,5 -30 -33 -60* -60 -63
Figure 1 Spectrum due to modulation and wideband noise measurement procedure
2.2. Spectrum due to Switching Spectrum due to switching is the second measurement defined with adjacent channel power. RF power quickly ramps up in GSM transmitters. This process of ramping should occur precisely and at a specific speed. Spectrum due to switching measurement checks for undesirable spectral components resulting from very quick power ramp. Quick power ramping causes significant interference in adjacent channels. This measurement usually detects faults in a transmitter’s output power amplifier. For the transmitter under test, all timeslots shall be set up to transmit full power. Similar to spectrum due to modulation, if SFH is on, any carrier may be used. Otherwise, only the carrier transmitted shall be used. At 400, 600, 1200, and 1800 kHz offsets, power shall be continuously measured using a video bandwidth of 100 kHz. The power measured shall not exceed the limits shown in Table 2 for a BTS [5] [6]. Table 2 BTS limits for spectrum due to switching
Offset (kHz) 400 600 1200 1800
GSM 900 Power (dBc) -57 -67 -74 -74
DCS 1800 Power (dBc) -50 -58 -66 -66
3. Modulation Accuracy Modulation accuracy test is characterized by phase error and frequency error sub-measurements. It reflects the performance of the transmitter; a significant phase error indicates a problem with the I/Q base-band generator, the Gaussian LPF, the modulator, or the RF amplifier of the transmitter. On the other hand, a significant frequency error indicates a problem with the synthesizer (phase-lock loop) [4].
Figure 2 Modulation accuracy test procedure Both sub-measurements depend on bits obtained after demodulation and are performed when all carriers are transmitting full power in all their time slots. In case of SFH, the BTS shall be hopping over the maximum number of carriers or else the test shall be only performed over the B, M, and T channels [5]. As defined in CCITT Recommendation O.153, any 148-bit subsequence of the 511-bit pseudo random sequence can be used to trace the trajectory of the expected phase. This expected phase is subtracted from the actual phase of the measured waveform. Phase error is determined using the root mean square (RMS) and the peak of the variation of the subtraction result, as shown in Figure 2. RMS and peak values should not surpass 5o and 20o respectively. Finally, the mean gradient of the subtraction result constitutes the frequency error. Frequency error should not exceed 0.05 ppm [5].
4. Mean Transmitted RF Carrier Power GSM BTS and MS are classified into classes according to the maximum power they can emit. Mean transmitted RF carrier power shall always be less than that maximum level. Were it not the case, transmitter power sources and amplifiers might be malfunctioning. Transmitted RF carrier power shall be measured at the input of the TX combiner. It is defined to be the mean power of the useful part of a GSM burst, shown in Figure 3. It keeps changing due to dynamic power control of the BTS in a predefined power steps defined by 3GPP TS 11.20 [5].
broadband down-converter that employs vector RF measurements. It has an 80 dB spurious-free dynamic range with a 30 dBm full scale input range. Correspondingly, the PXI-5620 is a high-spectral-purity single-channel digitizer module with a sampling rate ranging from 1 kS/s to 64 MS/s. It is characterized by an outstanding distortion-free performance due to its deep segmented memory and 14-bit resolution. A GSM 7 dBi directional antenna is used with the PXI-5660 via its SMA interface. Subsequently, the PXI5660 is connected to a PC via a PCI-PCI bridge (NI PXI8335), fiber optical cables, and a PCI card [3]. Figure 3 Mask limits of a GSM burst for BTS
5. Transmitted RF Carrier Power versus Time Transmitted RF carrier power versus time measurement evaluates the envelope of carrier power in the time domain within a predefined mask. To prevent interference, the power of GSM transmitters must ramp up then down within the allocated timeslot. If transmitters ramp up too slowly, data at the beginning of the burst may be lost. Furthermore, if transmitters ramp down too slowly, the user of the next timeslot will experience interference. A problem with the unit’s output amplifier is highly probable if the transmitter fails this measurement, which is performed using an analyzer in a zero-span mode. An example of the mask used in transmitted RF carrier power for BTS is shown in Figure 3 [4].
6. Implementation The above four measurements are implemented using LabVIEW, with Spectral Measurements Toolset and Modulation Toolkit, and included in the GMT. LabVIEW, a data-flow programming platform, provides a great graphical development environment for signal acquisition, measurement analysis, and data presentation. It delivers the flexibility of a programming language and avoids the complexity of traditional development tools. The basic unit in LabVIEW is the VI. The VI is an instrument driver divided into a front panel and a block diagram. The front panel forms the Graphical User Interface (GUI) of the driver while the block diagram forms graphical code which is compiled into machine code. Our GMT is designed for use along with NI PXI-5660 RFSA for GSM RF equipment testing. All its measurements are integrated in one VI.
6.1. RFSA Hardware The RFSA used for acquisition is the NI PXI-5660, which is a modular signal analyzer optimized for automated RF tests. PXI-5660 features a wide real-time bandwidth, a highly stable time base, and flexible software tools that can solve measurement applications ranging from component characterization in R&D to the remote monitoring of RF navigation systems. The main components of PXI-5660 are the NI PXI-5600 and PXI-5620. The PXI-5600 is a 2.7 GHz
As illustrated in Figure 4, a GSM signal passes through a range of stages while processed and analyzed by the PXI5660/GMT package.
Figure 4 PXI-5660/GMT operational overview First, the antenna detects required signal and passes it to the PXI-5600 down-converter where it is down-converted to an Intermediate Frequency (IF) of 15 MHz. Thereon, the signal is appropriately sampled and digitized by the PXI5620 digitizer. The sampling rate is carefully set in the GMT block diagram while the digitization is performed using 214 discrete levels. After the RF signal becomes digital, it is transmitted to the PC where users may carry out different tests using the GMT.
6.2. GMT Main VI All GMT measurements share the same block diagram. The general flow of each algorithm includes tuning to one of the 124 frequency carriers, synchronizing in time to a specific time slot, and finally isolating a number of GSM bursts. The process of tuning to a certain frequency f1 and acquiring a GSM signal is done through several VI’s. These VI’s are responsible for initializing the acquisition hardware and then configuring a filter with the required bandwidth. The filter bandwidth depends on the type of measurement undertaken. The ORFS measurement requires a bandwidth of 3.6MHz to ensure that all the offset frequencies are present for later processing. The other 3 measurements need a bandwidth of 200 kHz only, which is the effective bandwidth of a single GSM channel. Synchronization in time is achieved through the identification of the Frequency Burst, which has the shape of an unmodulated carrier in the frequency domain, and
then the deciphering of the following Synchronization Burst. The last stage of isolating a burst is simply done by saving the number of samples constituting the burst. The number of samples/burst depends on the sampling frequency used by the acquisition hardware [3].
that the MS doesn’t emit undesirable spectral components that cause inference in adjacent channels.
After employing the above procedures, specialized VI’s are used to perform measurement-specific operations and signal processing, such as filtering, power averaging, power spectrum calculations and GUI functions.
7. RF Test Results The developed GSM Measurement Toolkit is used to test the RF equipment, BTS and MS. For the ORFS measurement, the BTS failed the test while the MS passed it as shown in the two figures below. Figure 7 ORFS due to switching-MS (Passed) Over the useful part of the burst, as shown in Figure 8, the mean transmitted RF carrier power for a BTS is measured to be -58.6 dBm.
Figure 5 ORFS due to modulation-BTS (Failed)
Figure 8 Mean Transmitted RF Carrier Power-BTS The transmitted carrier power versus time measurement is conducted on both the BTS and MS as shown below in Figures 9 and 10.
Figure 6 ORFS due to modulation-MS (Passed) The BTS, far from the test location, failed to conform to the 3GPP standard due to the highly attenuated signal. On the other hand, the MS under test, which was placed near the antenna, passed the test due to limited effect of signal attenuation. The figure below also shows an MS that passed successfully the ORFS due to switching test. This indicates
Figure 9 Transmitted carrier power versus time-BTS (Passed)
engineers, particularly Mr. Mohamad Zeidan. His technical help was vital to the progress of our work. Finally, we value the assistance of the radio departments at France Telecom Mobile International, Ericsson, and LibanCell.
References [1] [2] Figure 10 Transmitted carrier power versus time-MS (Passed) The above two figures indicate that the GSM bursts, obtained from the downlink and uplink, fall within the limits of the measurement mask. This result guarantees that the power of both transmitters (BTS and MS) ramps up then down within the allocated timeslot causing no interference. Finally, the modulation accuracy test is performed on the MS over RF channel 22 (894.4M). As shown in Figure 11, the RMS and peak values of the differential phase graph are 1.355o and -4.160o, while the resulting frequency error is 13.6 Hz (-0.0152 ppm). Since the RMS and peak phase errors and frequency error are below the limits (5o, 20o, 0.05 ppm), the MS has passed the modulation accuracy test.
Figure 11 Modulation Accuracy Test-MS (Passed)
8. Conclusion In this work, we presented four basic tests performed on GSM RF equipment and the algorithms used to implement them. We also discussed the GSM Measurement Toolkit, which we developed for National Instruments using LabVIEW and PXI-5660. The importance of this toolkit lies in the fact that it offers the same functions provided by highly expensive GSM testing equipment, which are widely used by vendors and operators. Finally, we conducted realtime RF tests on a BTS and an MS.
Acknowledgements We would like to thank Prof. Ayman Kayssi for his supervision. His comments were enormously helpful. Also we deeply appreciate the support of National Instruments
[3] [4]
[5] [6]
Redl H., Weber K., Oliphant W. Introduction to Global System for Mobile Communications. Boston: Alba House, 1995. Eberspacher J., Vogel H. (1999). GSM: Switching, Services and Protocols. John Wiley & Sons, 1999. National Instruments Website [On-Line]. Available: http://www.ni.com Agilent Technologies. Understanding GSM/EDGE Transmitter and Receiver Stations and Their Components, Application Note 1312. Agilent Technologies, Inc. 2002. 3GPP Standards. TS 05.01, 05.02, 05.04, 05.05, 11.20, and 11.21. [On-Line]. Available: http://www.3gpp.org/specs/numbering.htm Mendlovitz, Mark A., “Reduction of Adjacent Channel Interference Effects in GMSK Communication Links”, IEEE Communications Magazine (2002)
