CHANNEL PROFILER AND METHOD OF PROFILING AN INCOMING

SIGNAL

FIELD OF THE INVENTION

The present invention relates to profiling an incoming signal, and more particularly to a channel profiler and a method of profiling an incoming signal.

BACKGROUND OF THE INVENTION In the last decades, the interest toward wireless communications has greatly increased. Such an interest has pushed the development and refinement of wireless protocols and technologies. All types of wireless communications have one thing in common: they allow data transmission over the air. However, transmitting data over the air introduces issues such as interference, distortion and multipath. To overcome such issues, multiple techniques for treating received data signals have been developed in combination with more robust modulation techniques. Some of the mostly used modulation techniques include Orthogonal Frequency Division Multiplexing (OFDM) and Orthogonal Frequency Division Multiplexing Access (OFDMA).

More particularly, OFDM and OFDMA modulation techniques include embedded mechanisms to reduce the effects of multipath. In these modulation techniques, the concept of guard interval has been introduced. The guard interval is located at the beginning of the OFDM symbol, and its length can be adjusted to overcome the effects of distance between two transmitters. The guard interval contains a cyclic prefix, which corresponds to a repeat of the end of the corresponding OFDM symbol copied into the guard interval. Although the length of the guard interval can be adjusted to optimize its impact on the effects of multipath, most standards set the length of the guard interval so as to increase interoperability. Thus, multipath and intersymbol interference remain important issues in standards relying on OFDM and OFDMA modulation techniques.

Many prior art documents propose methods and apparatuses for improving the quality of data received over the air using modulation techniques such as OFDM and OFDMA. Such documents generally propose techniques for profiling the incoming signal so as to perform appropriate synchronization, adjustment and correction. Typically, such profiling is performed while the incoming signal is being decoded.

United States Published Application 2004/0105512, published on June 3 ^rd , 2004, and assigned to Nokia Corporation describes a two-step procedure for OFDM receivers. The first step of the procedure relies on an autocorrelation of the received data signal so as to locate cyclic prefixes. The second step uses averaging of correlation results on a certain number of symbols to improve precision of synchronization. Although this invention describes some improvement, it is limited to the results of the averaging, which in some cases, is not good enough. Another document of interest on this topic is the article titled "Timing Recovery for OFDM transmission" published in the IEEE Journal on selected areas in communications, volume 18, number 11, in November 2000 by Baoguo Yang et al. This article also describes a two-step method. The first step of this method relies on autocorrelation of the received data, and more particularly the cyclic prefixes. Once the cyclic prefixes have been located, a second step is performed in the frequency domain. Performing profiling in the frequency domain is more complex, adds latency to the decoding process, and typically requires more power, which is not interesting for wireless applications, such as, for example, Wireless Broadband (WiBro). United States Patent No. 6,959,050, granted on October 25 ^th , 2005 to Motorola, lnc describes a method and an apparatus for profiling while synchronizing an OFDM signal in time, frequency and per-subcarrier rotation. More particularly, that patent also describes a two-step process, in which the first step consists of performing symbol timing synchronization and fractional frequency synchronization in the time domain. Afterwards, the second step of the process proceeds with performing per-subcarrier rotation synchronization

in the frequency domain. Although this patent describes an interesting solution, the proposed solution still relies on the frequency domain for performing per-subcarrier rotation synchronization so as to reduce multipath and intersymbol interference. The frequency domain is more complex than the time domain, and from a hardware perspective, also more power consuming. It is thus not an acceptable solution for wireless applications such as WiBro or IEEE Std 802.16.

There is therefore a need for a method and an apparatus for performing channel profiling efficiently, without adding any latency or requiring much power.

SUMMARY OF THE INVENTION

The present invention provides a channel profiler and a method for profiling an incoming signal. The channel profiler and method of the present invention rely on simpler methodology than those of prior art solutions, while reducing latency and power consumption issues.

For doing so, the method of the present invention includes a step of identifying a preamble of the incoming signal. The method pursues with extracting a reference waveform corresponding to the identified preamble and crosscorrelating the extracted waveform with the incoming signal.

In another aspect, the present invention is directed to a channel profiler. The channel profiler comprises an autocorrelator for identifying a preamble of an incoming signal. The channel profiler also includes a Discrete Fourier Transform module for providing reference waveforms and a crosscorrelator for crosscorrelating the preamble with a corresponding reference waveform.

BRIEF DESCRIPTION OF THE DRAWINGS

In the following description, the following drawings are used to describe and exemplify the present invention:

- A -

Figures 1a and 1b are a flowchart of a method for profiling an incoming signal in accordance with an aspect of the present invention;

Figure 2 is a block diagram of a channel profiler in accordance with an aspect of the present invention; Figure 3 is a graph depicting exemplary results of the autocorrelating of the method and channel profiler of the present invention;

Figure 4 is a graph depicting exemplary results of the crosscorrelating of the method and channel profiler of the present invention;

Figures 5a, 5b and 5c represent graphs of exemplary results of the summing of the crosscorrelation in accordance with the method and channel profiler of the present invention; and

Figure 6 is a schematic representation of the signal and its echoes in time domain.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to a channel profiler and a method for profiling an incoming signal. The present invention is characterized by many advantages over prior art solutions. First and foremost, the present invention provides a solution to prior art solutions which typically cause additional latency while profiling an incoming signal. For doing so, the present invention provides a methodology to perform channel profiling while the incoming signal is being processed by the transceiver, in the time domain.

As wireless transmissions cause some degradation on the signal, it is important to profile the incoming signal so as to apply necessary corrections to compensate for the incurred degradation. Typically, in prior art solutions, profiling is performed in the frequency domain. Once the incoming signal is transformed by use of a Discrete Fourier Transform module to a frequency domain, particularities of the incoming signal are visually recognizable. However, the time required to transform the incoming signal in the frequency domain causes some important latency, and requires increased power

consumption. Although latency and increased power consumption are acceptable trade-offs,, in some applications, for example Wireless Broadband (WiBro) systems, such trade-offs are not possible.

Reference is now made to Figures 1a and 1b, which depict, in accordance with the present invention, a method for profiling an incoming signal in the time domain. The method 100 is divided in 6 main sections: a section for identifying 105 a preamble, a section for extracting 110 a reference waveform, a section for crosscorrelating 115 the incoming signal with the extracted reference waveform, a section for summing results 120 so as to reduce the effects of multipath, a section for identifying 125 a maximum peak and a section for computing 130 a back-off.

The first section of the method relates to the identifying of the preamble. For doing so, any known method, such as autocorrelating the incoming signal, could be used. Alternatively, to obtain more detailed results, the identifying of the preamble could be performed in four sub-steps: autocorrelating 135 the incoming signal, locating 140 a plateau, autocorrelating a part of the plateau 145, and locating a peak on the obtained curve 150.

For example, in the case of an OFDMA system like WiBro, as a frequency pattern of the preamble is only populated with one third of sub-carriers, there is a (N/3)-pattern repetition (for a size N of DFT = 1024) in the time domain. This (N/3)-pattern repetition can be identified by an autocorrelation (eq. 1 and 2).

N /3-1 a(n) = y s(n + k)x s(n + k + N/3) ^* (1)

This autocorrelation 135 results in a plateau of 682 samples within the preamble zone, shown on Figure 3. This window contains a synchronization

point, which can be used, once identified, to train in an optimized manner the Discrete Fourier Transform module used to decode the incoming signal. Without prior knowledge of the synchronization point, the Discrete Fourier Transform module is trained blindly, i.e. sub-optimally. It is therefore very important to identify the synchronization point with accuracy. For doing so, the method of the present invention pursues by locating the plateau 140 in the results of the autocorrelation of the incoming signal. More particularly, in modulation techniques such as OFDM and OFDMA, symbols contain a cyclic prefix (CP), which is a copy of the last M samples of the symbol put in front of the symbol. Thus, a good estimate of the synchronization point can be given by a second autocorrelation on the cyclic prefix (Eq. 3 and 4).

M-\ a2(ri) = Y s(n + k) x s(n + k + N) ^* (3)

adliή) [A)

One can consider the WIBRO application as an example. In that technology, N=1024, N/3=341 and M=128. For a perfect channel, the two autocorrelations outputs of the incoming signal are represented in Figure 3, where the curve with the plateau represents the results of the first autocorrelation 135 and the other curve represents the results of the second autocorrelating 145. Thus, one can clearly see the window of (2N/3 samples, here 682) given by the first autocorrelation as well as the accurate synchronization point given by the peak index of the second autocorrelation, which corresponds to a beginning of the preamble. This first section of the profiling method is quite precise especially against important frequency offsets as only autocorrelations are used.

The method pursues with the second section of extracting the reference waveform 110. The extracting of the reference waveform 110 could be performed in alternate manners. For example, reference waveforms

representing all possible preambles in the frequency domain and/or the time domain could be stored in memory. Once the preamble has been identified in section 105, the corresponding reference waveform can then be identified. So the extracting section 110 accesses stored waveforms, in step 155, and a corresponding reference waveform is identified in step 160. In the event that the stored waveforms are in the frequency domain, as the incoming signal is still in the time domain, the identified reference waveform is transformed in the time domain in step 165, using for example an Inverse Discrete Fourier Transform module. Since the Inverse Discrete Fourier Transform module is already present in OFDM systems, and because the transformation of the reference waveform from the frequency domain to the time domain is only performed once, this aspect of the invention is quite advantageous as it does not add latency to normal operations of OFDM systems.

The method continues with section 115 by crosscorrelating the identified reference waveform in time domain, with the identified preamble (eq. 5 and 6).

N/3-1 c(ri) - Y s(n + k) x r(Jc) ^* (5)

For example, with the OFDMA modulation, as there is a 3x-repetition of the same pattern in the preamble, the result for the crosscorrelation of step 115 will correspond to the graph shown on Figure 4.

The method then proceeds with the summing 120 of the results of the crosscorrelating step 115, for which results are depicted on Figures 5a, 5b and 5c. The three peaks of Figures 5a, 5b and 5c correspond to 3 different paths with amplitudes proportional to a power of the path. For example, in Figure 5a, it is the second peak that has the greatest amplitude, while in

Figure 5b it is the third peak and in Figure 5c it is the first peak. Generally, the path with the greatest amplitude is chosen as the best synchronization point, but there are other factors, which could impact the selection of the best synchronization point. Therefore, the method continues with the section for identifying the optimal peak 125.

For example, in multipath environments, the choice of the synchronization point is crucial in order to minimize intersymbol and intercarrier interferences. It also has an impact on the interpolation performances for computation of a correction vector. Reference is now concurrently made to Figure 6, which depicts the incoming signal in multipath environment in the time domain. In Figure 6, each of the three signals corresponds to one of the peaks of the results of the crosscorrelation. By choosing the second crosscorrelation peak, the synchronization point obtained is not within the cyclic prefix of the first signal anymore, which will cause intersymbol intereferences. This analysis can be generalized to the other echoes. However, by selecting the first signal (i.e. the first peak), the synchronization point falls within the cyclic prefix for the 3 replicas of the signal. Therefore, to reduce intersymbol and intercarrier interferences, the method proceeds with the computing of a threshold value. The threshold value could advantageously correspond to an amplitude of the maximum peak of the crosscorrelation, divided by a constant, for example k. The constant k could be set to different values, to correspond to various environments. For example, in the case of the urban and rural environments, the k parameter could vary greatly. The method then pursues with the identifying of peaks above the threshold, step 175. Typically, the identifying of peaks will be performed in a set window. Then, the method proceeds with selecting one of the peaks above the threshold, in step 180. In a preferred manner, the first peak above the threshold is selected as the synchronization point. In the particular case where the delay of the main signal is higher than the cyclic prefix, the main peak is selected as the synchronization point. Indeed, in this case, the choice of the first peak would add intersymbol interference and intercarrier interference. By identifying the synchronization point in the propose manner,

more weight is given to the possible synchronization points that keep all signal replicas in the cyclic prefix. Additional flexibility is also provided by the present method in the even that the maximum peak is much stronger than the other peaks as intersymbol interference and intercarrier interference added by the other signal replicas will be limited.

Finally, the method ends with the step 130 of computing a back-off. The back-off corresponds to the difference between the selected synchronization point and an index of the peak with the greatest amplitude. For example, if the synchronization point selected corresponds to the first peak, which is above the threshold, but the first peak is not the peak with the greatest amplitude, which is the second peak, the back-off will correspond to 16 samples. As the synchronization point may not correspond to the first peak, the selection of the synchronization point may introduce some important phase rotation. The back-off can be used to improve performance of different blocks of a receiver, such as for channel interpolation, decoding, ... More particularly, for channel interpolation, the back-off can be used to compute a phase rotation pre-correction value and a phase rotation post-correction value. The phase rotation pre-correction value can be calculated with the following formula: ^L P ^C ~ ^{Lp x e}

θ = dx 2π/N where: C _pc is a phase rotation pre-correction value for pilot p;

Cp is an original correction vector for pilot p; d is the back-off;

N is the DFT size; and p is an index of the pilot relative to Direct Current.

After the channel interpolation has been performed, the correction vector is rotated as follows:

where: Crm is the phase rotation post-correction value for pilot p; Cm is a final correction vector for equalization; and m is a subcarrier index relative to Direct Current.

Thus the profiling method of the present invention provides a simple and efficient method of identifying a synchronization point, which takes under consideration effects of multipath, while reducing intersymbol and intercarrier interferences.

The present invention also describes a channel profiler for profiling the incoming signal. The channel profiler of the present invention is depicted on Figure 2. Generally, the channel profiler 200 includes an autocorrelator 210, a crosscorrelator 220, an Inverse Discrete Fourier Transform module 230, a summing module 240 and an identifying module 250.

The incoming signal 205 is received at the autocorrelator 210. The autocorrelator autocorrelates the incoming signal 205 with itself so as to identify the preamble thereof. Alternatively, the autocorrelator 210 may further proceed with locating a plateau of the autocorrelated incoming signal, autocorrelating a part of the plateau, and locating a peak of the autocorrelated part of the plateau. The located peak corresponds to a beginning of the preamble. The autocorrelator 210 provides the information of the preamble to the crosscorrelator 220.

The crosscorrelator 220 performs a crosscorrelation of the identified preamble with a corresponding reference waveform. The crosscorrelator 220 may either extract the reference waveform from the Inverse Discrete Fourier

Transform module 230, or extract it from a memory (not shown). The stored reference waveforms may either be in the time domain format, or in the frequency domain format. If the stored reference waveform is in the frequency domain, the Inverse Discrete Fourier Transform module 230 may further be used to transform the reference waveform in the time domain.

The crosscorrelated signal is then provided to the summing module 240 for summing. The summing module 240 may also be embedded within the crosscorrelator 220. The summed results is provided to the identifying module 250 either through the crosscorrelator 220, or directly (not shown).

Alternatively, the identifying module 250 could also be embedded in the crosscorrelator 220.

The crosscorrelator 220 may further include processing capabilities for computing a threshold value and for identifying all peaks of the summed results with an amplitude above the threshold value. Furthermore, the crosscorrelator 220 may further select a first peak above the threshold as synchronization point. The crosscorrelator 220 may also compute the backoff of the incoming signal, as previously described.

The present invention has been described by way of preferred embodiments. It should be clear to those skilled in the art that the described preferred embodiments are for exemplary purposes only, and should not be interpreted to limit the scope of the present invention. The method and channel profiler as described in the description of preferred embodiments can be modified without departing from the scope of the present invention. The scope of the present invention should be defined by reference to the appended claims, which clearly delimit the protection sought.