With more and more European operators, and not only those in the Eastern and Central European markets, introducing EDGE networks, Matthias Webe and Petri Toivonen say it is important to understand the additional tests that will be required to assure conformance of EDGE phones. But,they warn, testing phones to the minimum will be important to reduce measurement time.
EDGE or “Enhanced Data rates for Global Evolution” is a technology that was specified in 1997, but so far has had difficulties in making it to the market. There has always been the discussion what should come first, WCDMA (UMTS) or EDGE. How should EDGE be positioned — as an alternative to WCDMA or as a complement? In the US, EDGE was meant to pave the way towards 3G for the TDMA operators, where it was adopted in a slightly different way as “EDGE Compact” in order to the lower bandwidth available for mixed TDMA/EDGE networks. But this became obsolete when the TDMA operators decided to get rid of TDMA altogether in favour of GSM. Now even in Europe operators are seeing the benefit of EDGE and are implementing it. One mobile phone manufacturer has started to deliver EDGE-capable phones to the European, Asian and US markets, which makes it obvious that we eventually need to take a look at where does EDGE differ and what needs to be measured in addition to GSM, since EDGE phones will still be using the GSM mode for voice calls!
It should be stressed though that there will be a focus on the minimum tests required, rather than what can be tested, since especially in a service and manufacturing environment test time is crucial.
What is new with EDGE?
EDGE adds a new modulation scheme to the data protocols of both GPRS (General Packet Radio Service) and HSCSD (High Speed Circuit Switched Data), basically allowing EDGE to triple the data rates for these data services. When EDGE is applied to GPRS, it will be referred to as EGPRS (Enhanced GPRS), while HSCSD becomes ECSD (Enhanced Circuit Switched Data). However, since HSCSD is not very widespread this article concentrates on EDGE and GPRS. EDGE offers a choice between two modulation methods: GMSK and 8PSK. The latter transmits 3 bits per symbol, in contrast to GMSK modulation where only 1 bit per symbol is transmitted.
This means that the introduction of EDGE has the biggest impact on Layer 1 or the physical layer; it has only little impact on the upper layers. In the network, EDGE parameters only affect the base station and not the core network. Even though the impact for mobile phones is also mainly on the physical layer, this impact is much higher. The amplitude of the signals transmitted by the mobile phone is no longer constant because it is not rotating on a circle as it does for GMSK. It can actually change from one phase state to any of the eight phase states, having the effect that the amplitude may cross near the origin of the I/Q plane. (In order to avoid this, the whole I/Q plane is rotated by 22.5Â° or 3p/8 per symbol, which ensures that the signal never reaches the origin.)
This effect has an impact on the type of amplifier that can be used and may result in separate TX paths for GMSK and 8PSK. Because of the type of amplifier being used, the battery life will be constrained as well.
Like GPRS, EGPRS uses different coding schemes, combined with different modulation: GMSK or 8PSK. This allows operators to maintain the connection at reduced data rates under bad radio conditions. The table opposite (right) shows the different Modulation and Coding Schemes (MCS) for EDGE.
Is phone design affected?
The new modulation requires a completely new transmitter design. Due to the fact that the envelope is no longer constant, and therefore the amplitude is changing, for EDGE the power amplifiers (PAs) used are different. For this simple reason a common approach for RF designers will be that the transmit path is actually split into two: a path for GSM (GMSK) transmission and one for EDGE (8PSK) transmission. It is interesting to note though that the physical receive path remains the same. The receiver architecture is able to cope with the different types of modulation without the need to have two separate paths.
Other parts of the mobile phone that need modification include the demodulator, which needs to demodulate the signal without advance knowledge of the modulation type that the base station applies. This process is called blind demodulation. Also, the increased data rate implies that the channel codec must be of much improved capacity.
Since EDGE-capable phones still support GSM for voice, EDGE testing increases the overall test time. Therefore it is important to focus on those test criteria that are most significant for this new technology.
There is one RX path, which is usually tested as part of standard GSM testing. If the receiver quality is good for GSM it is also good for EDGE. If this were not the case there would be a flaw in the design of the receiver! Furthermore there are two TX paths, one for GMSK and one for 8PSK, which means that in addition to the standard GSM tests, the 8-PSK transmitter needs to be tested as well.
The tests foreseen for an EDGE transmitter are:
l Power measurements, incl. power-time-template
l Modulation quality measurements
One important quality factor with all RF systems is transmit power. The maximumpower level is key to a good data transfer rate in rural areas with a large cell radius. As the transmitter is in a different operating mode when applying 8PSK modulation, measurements are essential to ensure that the power level lies within the allowable limits. Note that the cell radius can be lower with 8PSK modulation than with GMSK due to the increased number of symbols.
To minimize unnecessary interference with adjacent cells and to ensure proper handover between cells, the power control mechanism in GSM and GPRS uses further power steps with exact levels. In manufacturing and after repair, these levels are usually calibrated with the help of a test set, with correction values being stored in the mobile phone.
Measurements of power versus time use the same principles for 8PSK as in plain GSM but with a modified template; the template takes into account that the 8PSK signal amplitude varies between symbols, depending on the exact symbol sequence. This attribute of 8PSK makes it more difficult to use cheap, nonlinear amplifiers and may lead to hardware and/or software designs correcting these nonlinearities. A power-time template for 8PSK signals shows a flat part at the beginning and the end of the burst applies to the tail bits, which have been chosen such that only minimal variations of the power level occur. The limits of the power level steps are identical to those known from GMSK modulation.
8PSK requires modulation measurements different from GMSK. The beauty of these measurements is that they provide parameters which are more likely to allow you to track down the source of any potential. The key measurement is that of the error vector magnitude or EVM for short, which is the distance in the I/Q diagram between the measured signal and the ideal one; the EVM is measured for each symbol separately. Standard measurement results to observe are the RMS-averaged EVM (for all the symbols of a burst), the maximum EVM within the burst, and the 95% EVM. The latter is the error vector magnitude not exceeded by 95% of the symbols within a burst; the parameter effectively disregards the highest 5% of all error vectors.
The error vector can be broken down into two components; one component is radial i.e. in the direction of the circle (phase error) and the other is vertical to the circle (magnitude error). These parameters are not part of the mandatory measurements but can be helpful in tracing a problem with the modulator.
Another parameter which is determined inside the test set but not necessarily shown or assessed is the amplitude change or droop, a measure of the stability of the power level during the burst. An amplitude change may be caused by on-chip temperature changes during the RF transmission.
Finally, if the I and/or the Q component of the modulator has an unwanted DC component, this can be determined in the origin offset suppression measurement. All these parameters, whether mandatory or not, are well known from the US-TDMA system which applies a slightly different modulation (p/4 DQPSK) but with very similar calculations of error parameters. — The table below summarises the measurement parameters with their test limits according to the ETSI standards.
A typical test scenario for a final test can be seen in Table 2. This table describes the test steps and their reasons.
In this table there are no RX tests on EDGE receivers for the above-mentioned reason. This of course only applies to a final test in a service or manufacturing environment. When performing type approval testing, EDGE receiver measurements are certainly an issue!
This article provided an overview of additional tests required in final testing in the manufacturing and service environment. It furthermore explained why it is acceptable to perform only TX measurements on EDGE radios. The purpose is to limit measurement time to the absolute minimum, rather than increasing measurement time unnecessarily. It is important to understand that there is still a need for testing GSM, since the 8PSK modulation introduced with EDGE is an add on rather than something exclusive.
Following this test philosophy will minimize the impact that EDGE has on testing time and at the same time ensure that the network quality is not compromised.