Orthogonal Frequency Division Multiplex OFDM is a special form of multi- carrier modulation that is arranged to provide high data rate communication by modulating data onto a plurality of sub-carriers. Wireless Local Area Network WLAN standards 802. 11 a and Hiperlan/2 are two examples of systems that have adopted OFDM modulation where for each system a data symbol is modulated onto 48 data sub-carriers with the use of 4 pilot sub-carriers to facilitate phase tracking to allow for coherent demodulation.

OFDM systems are arranged to provide several modulation and coding altematives, where each modulation and coding alternative allows different types of data mapping onto the respective data sub-carriers.

For example, seven physical layer modes are defined in the HiperLAN/2 specification ranging from a physical layer bit rate of 6 Mbps for Binary Phase Shift Key BPSK to 54 Mbps for 64 Quadrature Amplitude Modulation 64QAM.

During transmission and reception of an OFDM signal the signal will be attenuated and distorted by a frequency selective channel (i. e. the signals transmitted on the individual sub-carriers will be attenuated and distorted by different amounts dependent upon the environment where the channel characteristics are defined by a channel coefficient) in addition to noise and/or interference being added to the signal on each sub-carrier. The combination of noise and/or interference defines the quality of a communication channel (i. e. communication link) and is called the link quality.

The selection of a suitable modulation scheme is dependent upon the link quality of a communication link, such that if a selected data rate is not matched to the relevant link quality for a communication link data may either be lost, if the data rate selected is too fast for the given link quality, or a communication link may be under utilised.

However due to the variable environmental conditions of a wireless system the link quality of a communication link will fluctuate over a period of time.

Consequently, OFDM systems are arranged to dynamically select an appropriate modulation and coding rate parameter using a link adaptation procedure to dynamically measure link quality and selecting an appropriate modulation scheme.

One technique that is used for measuring the link quality is based upon frame error rate estimates that are obtained from cyclic redundancy checks CRC on decoded received data. This technique requires the reception of many frames of data to obtain reliable information, which for a communication link with rapidly changing link quality is undesirable.

Another technique for measuring the link quality, as described in US 2002/0110138 A1, is based upon a signal to noise ratio measurement that is performed on training symbols within the preamble, which are transmitted as part of a data burst on the data sub-carriers, and using data symbols that having been demodulated and decoded are re-encoded and re-modulated. This technique can be complex, requiring additional processing power, and can introduce additional errors. The implementation of the technique is furthermore not ideally suited for all applications and the determined link quality may not be optimal for all situations.

It is desirable to improve this situation and/or to provide alternative techniques for determining a link quality.

In accordance with a first aspect of the present invention there is provided a device according to claim 1.

This provides the advantage of allowing a link quality measurement to be made before the receipt of a whole frame and typically only requiring receipt of the preamble plus a few symbols before a good approximation of the link quality of a communication link can be performed. Further, this also provides the advantage of not requiring additional calculation to recode and remodulate a frame.

In accordance with a second aspect of the present invention there is provided a method for determining link quality of an OFDM communication link according to claim 10.

An embodiment of the invention will now be described, by way of example, with reference to the drawings, of which: Figure 1 shows an OFDM receiver according to an embodiment of the present invention; Figure 2 shows a OFDM media access frame; Figure 3 shows a signal to noise/interference ratio measurement unit according to an embodiment of the present invention.

The following embodiment is based upon a HIPERLAN/2 OFDM WLAN system, where a communication channel is defined as having a bandwidth of 20MHz, having 52 non-zero sub-carriers spaced 312.5 kHz apart, where the 52 sub-carriers consist of 48 data sub-carriers and 4 pilot sub-carriers. However, any

form of OFDM based system may be used that may have a different number of data and pilot sub-carriers.

Figure 1 shows an OFDM receiver 1. The OFDM receiver 1 includes an antenna 2 coupled to an RF front end module 3, the RF front end module 3 is coupled to a synchronisation unit 4, the synchronisation unit 4 is coupled to a demodulation unit 5, the demodulation unit 5 is coupled to a phase offset compensation unit 6 and an equalisation unit 7 with the equalisation unit 7 also being coupled to the phase offset compensation unit 6, the phase offset compensation unit 6 is coupled a data decoder 8. The phase offset compensation unit 6 and equalisation unit 7 are also coupled to a signal to noise and interference ratio measurement unit 9, according to an embodiment of the present invention, as described below. The signal to noise and interference ration measurement unit 9 is coupled to a link adaptation module 10.

The OFDM receiver 1 receives an OFDM signal over a communication channel and demodulates the time domain symbols into frequency domain sub- carrier via the demodulation unit 5 and decodes data received over this channel via the data decoder 8, as is well known to a person skilled in the art.

The equalisation unit 7 compensates for the attenuation due to the frequency selective channel by first measuring the channel coefficients during channel coefficient estimation, which consist in storing the non-zero sub-carrier information receiving during the preamble long symbol, and then compensating the data sub- carriers either by dividing them by the channel coefficients or by multiplying them by the complex conjugate of the channel coefficients if just the phase needs to be compensated, as is well known to a person skilled in the art. Hence, the channel coefficient may be seen as the complex sub-carrier symbol which is received in the sub-carrier if a data symbol of unity amplitude and zero phase offset was transmitted. Thus, the channel coefficient not only reflects the channel transfer

function but is also a direct amplitude and phase measure for the received symbol assuming that a constant transmit power is used.

The phase offset compensation unit 6 compensates for the frequency offset that exists between the transmitter and the receiver carrier central frequency.

OFDM transceivers are very sensitive to this frequency offset, a first frequency offset compensation is implemented in the time domain (in the synchronization unit 4). However a second phase compensation is implemented in the frequency domain in the phase error compensation unit 6 that is more refined and can be recalculated for each symbol. The phase error compensation unit 6 applies phase compensation to the data symbols by rotating each sub-carrier with the opposite of the phase error that occur on sub-carrier i for symbol k (i. e. ferror, i (k) 4) comp, i (k) =-+error, i (k)). The phase error is partly derived from the frequency offset which is measured by the synchronization unit 4 when the preamble is received.

The phase error compensation is implemented by multiplying each data sub- carrier by a complex number: Figure 2 illustrates a HIPERLAN/2 medium access control MAC transmission frame 200. The MAC frame 200 consists of a broadcast control channel BCCH 201 followed by a frame control channel FCC 202 followed by the respective downlink traffic 203 and uplink traffic 204 that is followed by a random access channel 205.

The data transmitted as part of the down link traffic 203 is contained within bursts 206 that consists of a preamble part 207 and protocol data units 208.

The preamble part 207 comprises a cyclic prefix and two training symbols (not shown), where the training symbols are used by the equalisation unit 7 to perform a channel coefficient estimation, as is well known to a person skilled in the art.

Figure 3 shows the signal to noise/interference ratio measurement unit 9.

The signal to noise and interference ratio measurement unit 9 includes a signai power estimator module 301, a pilot noise and interference power module 302 and a signal to noise/interference ratio calculator module 303.

The signal power estimator module 301 receives channel coefficient estimation information from the equalisation unit 7, the pilot noise and interference power module 302 receives phase information from the phase offset compensation unit 6 and the signal to noise/interference ratio calculator module 303 is coupled to the output of both the signal power estimator module 301 and the pilot noise and interference power module 302 to allow calculation of a signal to noise/interference ratio 303 for a received signal, as described below.

The signal to noise/interference ratio measurement unit 9 is arranged to calculate the SNIR, which is used to determine the link quality of a communication link, on a symbol by symbol basis (i. e. the SNIR is determined over the 52 sub- carriers).

For each received symbol the equalisation unit 7 calculates a channel coefficient estimation over each sub-carrier using training symbol information contained with the preamble of an OFDM burst. For the purposes of this embodiment, the channel coefficient estimation can be, for more accuracy, derived from the average of the two training symbols contained within the preamble.

Specifically, the equalisation unit 7 may calculate the channel coefficient Hi as the received sub-carrier symbol amplitude for a transmitted symbol of unity energy and zero phase offset. Hence, the squared channel coefficient Hj2 for each sub-carrier is indicative of the attenuation in the channel and also (for a constant transmit power) the energy of the received symbol.

Using the channel coefficient estimation H provided by the equalisation unit 7 the signal power estimator module 301 calculates the average signal power for the sub-carriers using the following equation: where i is used to denote the sub-carrier and (Hj) 2 is the squared channel coefficient for sub-carrier i equivalent to the energy of the symbol received in sub- carrier i. The equalisation unit 7 thus determines the average energy received for a given symbol and in particular this average energy is calculated for a training symbol for which the symbol values, and thus the transmitted symbol energy, are known and constant. Hence, S2 is an energy measure corresponding to the received averaged signal energy for a training symbol comprising 52 sub-carrier symbols.

The average signal energy value is provided to the signal to noise/interference ratio calculator module 303.

For each received non-training data data symbol k, which is received over the 48 data sub-carriers, the pilot noise and interference power module 302 calculates the average energy of the sub-carrier variations for the associated pilot information, transmitted over the four pilot sub-carriers, as described below.

Hence, for non training data, the pilot noise and interference power module 302 determines a noise and interference indication by evaluating the difference between the pilot data symbols received in the four pilot sub-carriers and the expected value in the absence of noise (including non-compensated channel variations) and interference.

First, the noise and interference NI ; (k) of a pilot signal associated with a received data symbol k is calculated by the pilot noise and interference power module 302 by subtracting the pilot sub-carrier Pi (k) received with the kth symbol, having been corrected for phase compensation, from the pilot sub-carrier channel coefficient estimate Hi, as given by the following equation: Hi is the channel coefficient determined by the channel estimation and thus corresponds to the sub-carrier symbol received in pilot sub-carrier i for the training data. It is assumed that the same sub-carrier symbol is transmitted in the same pilot sub-carrier for data symbol k. Accordingly, if no noise or channel variations occur between the transmission of the training data and data symbol k, it is to be expected that Hi is equal to Pi (k). However, a frequency error will result in a phase variation for data symbol k with respect to the training data and this phase variation is denoted 4) comp (k) and is estimated by the phase offset compensation unit 6. Hence, the factor e-j.#comp(k) compensates for the frequency error.

Accordingly, Pi (k)e-j#comp,f(k) corresponds to the expected sub-carrier symbol to be received for data symbol k in pilot sub-carrier i in the absence of noise (including unexpected channel fluctuations) and interference. Accordingly, NI ; (k) represents an indication of the noise and interference of the sub-carrier pilot symbol in pilot sub-carrier i.

Having calculated the noise and interference NI ; (k) of the individual pilot signals as described above, the pilot noise and interference power module 302

then calculates the average energy for the pilot signal noise and interference associated with the kth symbol using the following equation: ..-) -/ NI 2 (k) = i=PilOhnanber 4 i. e. by averaging the individual noise and interference estimates of the four individual pilot sub-carriers.

It can be assumed that the average energy for the pilot signal noise and interference will be the same as the average energy for the noise and interference of the whole associated kth symbol. Therefore, the value N12 (k) can be regarded as the average energy for the associated kth symbol noise and interference. Thus, an estimate of channel noise and interference (including channel phase noise and amplitude noise) is determined based on a comparison between the received pilot sub-carrier symbols and the sub-carrier symbols (channel coefficients H) obtained during the channel estimation.

The noise and interference energy value N 12 (k) is provided to the signal to noise/interference ratio calculator module 303.

Using the received values of S2 and N12 (k) the signal to noise/interference ratio calculator module 303 calculates a SNIR for the 0 symbol, in decibels, using the following equation: SNIR (k) =20log(S/NI (k) ) =1 Olog (S2) Olog (NI2 (k))

SNIR (k) = 1OlogtE (13) i=1 (H ;) 2)-10 logt (Hi (k) e~ k)) 2) _1 OlogHaving calculated a SNIR for a given symbol the signal to noise and interference calculator module 303 can also average the calculated SNIR for all the N symbols in a frame (or over a number of frames): The above description assumed that the pilot data symbols were constant for all symbols k. However, it will be appreciated that the approach can easily be adapted to take into account the use of different pilot data symbols in different symbols k. Thus, the expected symbol value may be determined in response to a transmit data value associated with the received symbol value. Specifically, the phase compensation term may be modified to include the phase difference between the pilot symbol of the training data and the pilot symbol of symbol k. For example, if the transmitted symbols have been scrambled by the transmitting node (e. g. by multiplication of a constant pilot symbol by a given sequence of 1 and-1), the above equations can be modified to include an appropriate descrambling function (i. e. Djfk)). Accordingly, the above equations can be modified as follows : NI (k) = H (k) D °^, M ) ,-) "'' NI2 4

SNIR (k) =lOlo (k) (13) i=l The signal to noise/interference ratio calculator module 303 provides the SNIR information to the link adaptation module 10. The link adaptation module 10 uses the SNlR information to determine an appropriate modulation mode for modulating data to be communicated over the communication link.

In order to perform optimum link adaptation (i. e. determination and selection of an appropriate modulation mode) it is desirable to combine the SNIR measurement with information about the channel type (e. g. delay spread).

One example of a link adaptation algorithm that can be used by the link adaptation module 10, for determining a suitable modulation mode, is described below. The link adaptation algorithm is based upon the assumption that the channel type, characterised by its delay spread, is stable and does not vary when the receiver 1 stays within the same environment and, as such, it is possible to combine the SNIR with some PER (Packet (or PDU-Packet Data Unit) Error Rate) measurement.

The algorithm comprises the following steps for a received frame: (i) Using simulation to obtaining a table containing the data rate as a function of the SNR for the default channel type (the most probable one) and stored this table in MAC software memory; (ii) For each received frame the PER is checked and if it is above PERMAX (a typical value can be 0.01) then a correction factor SNIF COR is incremented by 1 but if the PER is below a second treshold PER MIN (typically

0.001) it is decremented by 1 and if the PER is between these 2 tresholds, SNIF COR is not updated; (iii) Obtaining the optimum data rate by reading the data rate table with SNIR-SNIR_COR.

At the next frame, the same process is executed, if the PER is above PER MAX, the correction factor is increased (it can be incremented for each wrong frame to converge more quickly). After reaching a PER below PERMAX we know that an optimum correction value corresponding to the channel delay spread has been selected.

It will be apparent to those skilled in the art that the disclosed subject matter may be modified in numerous ways and may assume may embodiments other than the preferred forms specifically set out as described above, for example OFDM systems having any number of data and pilot sub-carriers could be used.