WO2013167739A1

WO2013167739A1 - Preamble detection in presence of carrier offset

Info

Publication number: WO2013167739A1
Application number: PCT/EP2013/059737
Authority: WO
Inventors: Robert Young; Brian Gaffney
Original assignee: Neul Ltd
Priority date: 2012-05-11
Filing date: 2013-05-10
Publication date: 2013-11-14
Also published as: GB2503869A; GB201208244D0; GB2503869B

Abstract

A method for detecting the start of a frame in a signal comprising cross correlating the signal with a known preamble sequence, conjugate multiplying the correlation results with a delayed version of it (delay equal to preamble length) in order to cancel the frequency offset effect of increasing the phase of consecutive correlation results, and finally coherently combining the phase corrected correlation results and compare with a detection threshold.

Description

PREAMBLE DETECTION IN PRESENCE OF CARRIER OFFSET

The invention relates to a method for determining the start of a communication at a receiver in a wireless communication system.

A communication may be arranged In the form of a frame comprising a preamble for synchronisation followed by a data payload. The preamble may comprise multiple repetitions of a synchronisation word that is known to the receiver. To detect the start of the frame, the receiver has to detect this repeated pattern in the presence of noise. The receiver typically does not have prior knowledge of the frequency error between it and the transmitter or of the channel impulse response. The benefits of accurate and fast detection of the start of the frame include allowing the system to reduce the length of the preamble required in each frame, allowing the receiver to dedicate more of the received preamble data to frequency or channel estimation and improving the overall detection rate (which includes increasing "detects" and reducing false alarms).

The goal of the preamble detection is to detect when a frame is present, i.e. when a channel exists at the output of the correlator. In one method, the received samples are initially correlated with the synchronisation word. An example output of this stage is shown in Figure 2, where z is the output of the correlator. Some lags will generate peaks in the correlator output. In Figure 2, a lag of 5 samples generates the highest peaks. The synchronisation word is L samples long, and therefore the peak repeats itself every L samples. Each of these peaks represents an estimate of the channel impulse response.

Figure 2 illustrates an example of a correlator output when there is no frequency error between the transmitter and the receiver and the channel does not change significantly over the length of the preamble. Consequently the peaks repeat themselves at a regular spacing of L samples. In practical systems there may be a frequency error between transmitter and receiver, which is unknown to the receiver. In this case the channel impulse response is still repeated, but with an unknown phase shift across the synchronisation words. To detect the presence of a preamble, one method takes the absolute value or power of z. Successive values for each lag are filtered, or "windowed", in order to emphasise recent measurements and discard older measurements. Two examples of windowing are exponential windowing (e.g. single tap MR), which reduces the influence of older measurements on the current output, and rectangular windowing, which limits the number of past measurements considered. The output of the windowing function is input to a decision device, which determines if a channel is present. For example, the maximum value at any lag may be compared to a static or adaptive threshold to determine if a channel exists. If the threshold is too low, detection can be triggered by noise. If it is too high, frames may be missed.

A downside of the existing method is that it discards phase information by taking the absolute value or power of the correlator output to non-coherently combine the correlator outputs for a specific lag. As a result, when a particular lag generates a channel impulse response, signal energy in that response is lost. It also causes an accumulation of non-zero mean noise power.

Therefore, an improved method for detecting a preamble in a received signal is required.

According to a first embodiment of the invention, there is provided a method for detecting the start of a frame in a signal, the method comprising comparing a first section of the signal with a known sequence that indicates the start of the frame to generate a first comparison including a first phase component, comparing a second section of the signal with the known sequence to generate a second comparison including a second phase component that incorporates an additional rotation relative to the first phase component and processing the second comparison with the first comparison to generate a combined comparison that is representative of said second comparison but without the additional rotation.

The method may comprise selecting the first and second sections of the signal to be representative of a potential starting point of the frame. The method may comprise selecting the first and second sections such that at least part of the second section is formed from a part of the signal received after a part of the signal that forms the first section.

The method may comprise selecting the first and second sections to comprise non- overlapping sections of the signal.

The method may comprise selecting the first and second sections to comprise a number of contiguous samples of the signal.

The method may comprise selecting the first and second sections to comprise a number of samples equal to the length of the known sequence.

The method may comprise selecting the first and second sections so that the first sample of the second section and the first sample of the first section are spaced in the signal by the length of the known sequence.

The method may comprise comparing the first and second sections with the known sequence to generate first and second comparisons that each include a phase component representative of a phase difference between the known sequence and the respective one of the first and second sections.

The method may comprise comparing the first and section sections with the known sequence to generate first and second comparisons that each include a component indicative of a gain associated with the channel over which the signal was received.

The method may comprise comparing the first and second sections with the known sequence by performing a correlation operation between each respective section and the known sequence.

The method may comprise processing the second comparison with the first comparison by multiplying the second comparison with the complex conjugate of the first comparison. The method may comprise, while the second section is being compared with the known sequence to generate the second comparison, delaying the first comparison and taking the complex conjugate of the delayed first comparison.

The method may comprise comparing a third section of the signal with the known sequence to generate a third comparison including a third phase component that incorporates an additional rotation relative to the second phase component, processing the third comparison with the second comparison to generate a combined comparison that is representative of said second comparison but without the additional rotation and coherently combining the combined comparisons generated from the second and third comparisons.

The start of the communication may be indicated by a repeated transmission of the known sequence, and The method may comprise selecting a potential starting point for the frame in the signal and dividing the received signal into consecutive, non- overlapping sections, the first section commencing at the potential starting point and each section being equal in length to the known sequence.

The method may comprise, for each section: comparing that section with the known sequence to generate a comparison including a phase component; and combining that comparison with the comparison generated for the preceding section to form a combined comparison that includes a combined phase component substantially the same as the phase component generated in respect of the first section.

The method may comprise combining all of the combined comparisons generated in respect of the non-overlapping sections of the signal.

The method may comprise coherently combining all of the combined comparisons.

The method may comprise combining ail of the combined comparisons by weighting combined comparisons generated in respect of more recently received sections of the signal more heavily than combined comparisons generated in respect of less recently received sections of the signal.

The method may comprise determining, from the combination of the combined comparisons, whether the potential starting point corresponds to the actual starting point of the frame.

The method may comprise determining that the potential starting point corresponds to the actual starting point of the frame if the combination of the combined comparisons exceeds a predetermined threshold.

The method may comprise selecting multiple different potential starting points for the frame, and for each potential starting point, dividing the received signal into consecutive, non-overlapping sections commencing at the potential starting point.

The method may comprise generating a combination of combined comparisons in respect of each starting point and determining the actual starting point of the communication in dependence on those combined comparisons.

The method may comprise determining an average of the combination of combined comparisons for all of the potential starting points and determining that a potential starting point is the actual starting point if the combination of combined comparisons in respect of that potential starting point exceeds the average by a predetermined amount.

According to a second embodiment of the invention, there is provided a receiving device configured to detect the start of a frame in a signal by comparing a first section of the signal with a known sequence that indicates the start of the frame to generate a first comparison including a first phase component, comparing a second section of the signal with the known sequence to generate a second comparison including a second phase component that incorporates an additional rotation relative to the first phase component and processing the second comparison with the first comparison to generate a combined comparison that is representative of said second comparison but without the additional rotation.

For a better understanding of the present invention, reference is made by way of example to the following figures, in which:

Figure 1 shows an example of a typical synchronisation preamble and the output from the cross correlation with each section S_kof the synchronisation preamble;

Figure 2 shows an example of processing blocks involved in a preamble detect scheme;

Figure 3 shows an example of a process for detecting a preamble in a received signal;

Figure 4 shows an example of a correlation operation; and

Figure 5 shows an example of a delay line and windowing stages.

The following description is presented to enable any person skilled in the art to make and use the system, and is provided in the context of a particular application. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art.

The general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

One embodiment relates to a method of detecting the start of frame in a wireless communication system, where the time of arrival of the start of the frame is not exactly known. The term "frame" refers to a communication comprising the preamble sequence followed by a data payload. The preamble is provided to aid the receiver with synchronisation. The preamble suitably includes multiple repetitions of a synchronisation sequence. The sequence is preferably known to the receiver. The sequence may be of length L chips.

There may be a time-varying phase difference between the signal and the known synchronisation sequence due to a frequency mismatch between a transmitter that transmitted the signal and a receiver that received the signal. Consequently, there is a phase difference between the received signal and the synchronisation sequence that varies across the signal. Therefore, a comparison operation that correctly lines up the known sequence with the received signal may generate an estimate of the channel impulse response that is rotated due to the transmitter-receiver frequency mismatch. If the signal includes a repetition of the synchronisation sequence, so that the comparison generates repeated estimates of the channel impulse response (e.g. in the way shown in Figure 1), each of those estimates will be rotated by a different amount. In other words, the channel impulse response is modulated by the frequency error between the transmitter and the receiver. One or more embodiments of the invention may address this problem by processing two or more outputs of the comparison operation with the aim of removing the additional rotation. The receiver does not require prior knowledge of the frequency error between transmitter and receiver or the channel impulse response to implement this method.

The method may comprise comparing sections of the received signal against the known synchronisation sequence to produce a comparison associated with that particular section of the signal. This may be achieved using a correlation operation or similar. The resulting comparison will typically comprise a complex number representing a channel gain and a phase component representing an unknown phase shift between that section of the signal and the synchronisation sequence. This unknown phase component will differ from section to section of the signal due to the time-varying nature of the phase shift.

The method may include removing the unknown frequency offset by combining the comparisons generated by two or more different segments of the signal. The additional rotations caused by the frequency mismatch are preferably subtracted during the processing so that each processed comparison has a phase component that is substantially the same. Having the same phase component across the comparisons allows them to be combined coherently, enabling more of the processing gain achievable from the preamble's repetitive structure to be retained.

The time-varying component of the phase difference could be either positive or negative. Suitably the frequency error between the transmitter and receiver is constant over the time frame of interest so that the time-varying component of the phase difference changes linearly with time.

The methods described herein allow for fast and accurate detection of frames at low signal to noise ratios. Coherently combining the received data reduces the overhead required to improve the signal-to-noise ratio. This improves the detection rate of a receiver by increasing the probability of frame detection at low signal to noise ratios. The probability of a false alarm is also reduced. These improvements can be achieved without the receiver requiring any knowledge of channel response or frequency error.

Examples of the processing blocks that might be involved in a preamble detection scheme are shown in Figure 2. One or more of these blocks may be comprised in a receiver or other device, and may implement the following:

• A cross correlation unit, where the received samples are correlated against a known preamble sequence.

• A delay line unit, where the output of the cross correlator is multiplied by the conjugate of the cross correlator output from L samples ago (where L is the length of the synchronisation word). The delay line is shown in more detail in Figure 4.

• A windowing unit, where the previous values for a potential lag are windowed and summed. If all potential lags are tested, and the length of the synchronisation word is L, this requires a total of L windowing blocks. Any form of windowing function might be used. Examples of suitable functions include exponential windows (which may be implemented by a single tap IIR filter) and moving average windows.

• A decision unit, where the outputs of the windowing stage for all lags are compared to some threshold. The threshold may be adaptive or static. A preamble may have been detected if a lag exceeds the threshold.

An example of a process for detecting the start of a frame Is shown in Figure 3. The process starts in step 301. In step 302 the signal is received. In step 303 the received signal Is correlated against the known synchronisation sequence. The output of this stage consists of noise when no frame is present, and a noisy estimate of the channel impulse response when a frame is present. The estimate of the channel impulse response is modulated by the frequency error between the transmitter and receiver.

An example of a cross correlation is shown in Figure 4, which shows the known synchronisation sequence 402 being moved sample-by-sample (403) along the received signal 401. In this example, a correlation output 404 is generated at each sample offset. The outputs separated by the length of the synchronisation sequence L (404,405) relate to the same potential lag. The outputs relating to the same lag will, however, each incorporate a different phase component due to the time-varying rotation introduced by the frequency error.

In step 304 the outputs relating to a particular lag are passed to a delay line. During this stage, the output of the correlator is multiplied by the conjugate of the output value L samples ago. An example of a suitable delay line is shown in Figure 5.

To gain insight into the delay operation, let the channel gain be represented by the complex number a and the phase shift due to frequency error between samples spaced L samples apart be represent by Θ. Consecutive correlator outputs for lag m can be written as follows: z_m =a exp(j9) + n_m

Zm_+L =a exp(j26) + r z_m+2L =σ exp0^'38) + n_m+2L (1) where n is a complex Gaussian noise component.

At time m+L, the current correlator output z_m+L is multiplied by the conjugate of the correlator output from time m, ¾: z_m+i Conj(z_m) = a exp(j2e)conj(a)exp(-jO) + n = la!² expQO) + n (2) where n is a noise term.

The phase angle θ is dependent on the frequency error only and the angle of the complex number a is cancelled by the conjugation.

At time m+2L, the correlator output z_m+2L is multiplied by the conjugate of z_m+L, resulting in the same expression (with a different noise term) as equation (2) above. The same process is repeated across all offsets and all potential lags.

Comparing equations (1) and (2) above it can be seen that the delay stage of the process generates a series of corrected correlator outputs that all have the same phase. The unknown phase component introduced by the time-varying rotation has been removed and all the corrected outputs have the same phase component. By rendering the phase of the signal term constant, this stage allows all subsequent stages to take advantage of the phase information.

Equations (1) and (2) define the combining of consecutive outputs of the correlator. While this is likely to be the most straightforward implementation, it is not the only option. For example, a similar result could be achieved in terms of equalising the phase component across the outputs if output 1 were combined with output 3 and output 2 with output 4. The resulting two combined outputs would incorporate substantially the same phase component (e.g. 2Θ in the notation of equations (1) and (2)) and could be combined coherently. A similar variation is possible by using overlapping sections of the signal between which the phase angle of the correlator outputs may vary by a fraction of the phase angle Θ seen between correlator outputs for non-overlapping sections of the signal.

In step 306 the method passes to a windowing stage that combines observations over multiple synchronisation words. The windowing stage may also introduce some processing gain where a signal is present. In one or more embodiments of the invention, the combining is done coherently, which allows the method to take advantage of the potential processing gain offered by the phase information. Preferably, the combining process emphasises recent measurements over older measurements. One example of a suitable windowing function is an exponentially weighted window function (which may be implemented using a single tap IIR) where the influence of older measurements on the output of the windowing stage decreases exponentially. An example of a possible implementation is illustrated in Figure 5, where exponential windowing is illustrated for the I^th lag. Another example of a suitable windowing function is a rectangular windowing function, where the previous K values are weighted equally but older values are discarded.

Step 307 is the detection stage. This stage determines if a preamble is present at a particular lag. There are a number of different ways detection might be triggered. One is for detection to be triggered if the power/absolute value of the windowing output for a lag exceeds a predetermined threshold. Another example compares the windowing outputs of different potential lags. For example, the jeak power/absolute value of each lag might be isolated and compared to the mean of the other lags. If the peak exceeds the mean by some multiple, then detection may be triggered.

The preamble detection methods described herein allow retention of phase information. The result is improved signal energy and zero mean noise accumulation. This allows for greater separation of noise and signal, which results in more efficient detection (since fewer synchronisation words are required to achieve detection) for high signal-to-noise ratios and more accurate detection (since the noise does not accumulate) for low signal-to-noise ratios. The apparatus shown in Figure 2 is shown illustratively as comprising a number of interconnected functional blocks. This is for illustrative purposes and is not intended to define a strict division between different parts of hardware on a chip. In practice, the apparatus preferably uses a microprocessor acting under software control for implementing the methods described herein. In some embodiments, the algorithms may be performed wholly or partly in hardware. In particular, the implementation example shown in Figure 5 may be put into practice using delay lines that are at least partly implemented in hardware.

An apparatus configured to implement some or all of the processes described herein may be wholly or partly comprised within a receiver. Alternatively, the apparatus may be implemented in a separate piece of equipment from a receiver and the receiver may pass the received signal, or a processed version of it, to the apparatus for processing.

The embodiments described herein might be advantageously implemented in any wireless communication network, he method might, for example, be advantageously implemented in a communication letwork that implements a communication protocol such as the Weightless protocol for machine communications. In Weightless, communication between a base station and its associated terminals is achieved by means of a series of frames, each comprising a preamble, a downlink data portion and an uplink data portion. The embodiments described herein might be implemented in terminals, base stations and/or any other network equipment. Weightless is also designed to operate in unlicensed parts of the frequency spectrum, which are particularly susceptible to low SNR conditions. Weightless is mentioned for the purposes of example only. It should be understood that the methods described herein might be implemented in accordance with any communication protocol in which communications incorporate synchronisation sequences as an aid to receiving devices and/or which operate in unlicensed parts of the frequency spectrum.

The applicants hereby disclose in isolation each individual feature described herein and any combination of two or more such features, to the extent that such features or combinations are capable of being carried out based on the present specification as a whole in light of the common general knowledge of a person skilled in the art, irrespective of whether such features or combinations of features solve any problems discloses herein, and without limitation to the scope of the claims. The applicants indicate that aspects of the present invention may consist of any such feature or combination of features. In view of the foregoing description it will be evident to a person skilled in the art that various modifications may be made within the scope of the invention.

Claims

1. A method for detecting the start of a frame in a signal, the method comprising: comparing a first section of the signal with a known sequence that indicates the start of the frame to generate a first comparison including a first phase component;

comparing a second section of the signal with the known sequence to generate a second comparison including a second phase component that incorporates an additional rotation relative to the first phase component; and

processing the second comparison with the first comparison to generate a combined comparison that is representative of said second comparison but without the additional rotation.

2. A method as claimed in claim 1 , comprising selecting the first and second sections of the signal to be representative of a potential starting point of the frame.

3. A method as claimed in claim 1 or 2, comprising selecting the first and second sections such that at least part of the second section is formed from a part of the signal received after a part of the signal that forms the first section.

4. A method as claimed in any preceding claim, comprising selecting the first and second sections to comprise non-overlapping sections of the signal.

5. A method as claimed in any preceding claim, comprising selecting the first and second sections to comprise a number of contiguous samples of the signal.

6. A method as claimed in claim 5, comprising selecting the first and second sections to comprise a number of samples equal to the length of the known sequence.

7. A method as claimed in claim 5 or 6, comprising selecting the first and second sections so that the first sample of the second section and the first sample of the first section are spaced in the signal by the length of the known sequence.

8. A method as claimed in any preceding claim, comprising comparing the first and second sections with the known sequence to generate first and second comparisons that each include a phase component representative of a phase difference between the known sequence and the respective one of the first and second sections.

9. A method as claimed in any preceding claim, comprising comparing the first and section sections with the known sequence to generate first and second comparisons that each include a component indicative of a gain associated with the channel over which the signal was received.

10. A method as claimed in any preceding claim, comprising comparing the first and second sections with the known sequence by performing a correlation operation between each respective section and the known sequence.

11. A method as claimed in any preceding claim, comprising processing the second comparison with the first comparison by multiplying the second comparison with the complex conjugate of the first comparison.

12. A method as claimed in claim 1 1 , comprising:

while the second section is being compared with the known sequence to generate the second comparison, delaying the first comparison; and

taking the complex conjugate of the delayed first comparison.

13. A method as claimed in any preceding claim, comprising:

comparing a third section of the signal with the known sequence to generate a third comparison including a third phase component that incorporates an additional rotation relative to the second phase component;

processing the third comparison with the second comparison to generate a combined comparison that is representative of said second comparison but without the additional rotation; and coherently combining the combined comparisons generated from the second and third comparisons.

14. A method as claimed in any of claims 2 to 13, in which the start of the communication is indicated by a repeated transmission of the known sequence, the method comprising;

selecting a potential starting point for the frame in the signal; and

dividing the received signal into consecutive, non-overlapping sections, the first section commencing at the potential starting point and each section being equal in length to the known sequence.

15. A method as claimed in claim 14, the method comprising, for each section: comparing that section with the known sequence to generate a comparison including a phase component; and

combining that comparison with the comparison generated for the preceding section to form a combined comparison that includes a combined phase component substantially the same as the phase component generated in respect of the first section.

16. A method as claimed in claim 15, comprising combining all of the combined comparisons generated in respect of the non-overlapping sections of the signal.

17. A method as claimed in claim 16, comprising coherently combining ail of the combined comparisons.

18. A method as claimed in claim 16 or 17, comprising combining all of the combined comparisons by weighting combined comparisons generated in respect of more recently received sections of the signal more heavily than combined comparisons generated in respect of less recently received sections of the signal.

19. A method as claimed in any of claims 16 to 18, comprising determining, from the combination of the combined comparisons, whether the potential starting point corresponds to the actual starting point of the frame.

20. A method as claimed in claim 18, comprising determining that the potential starting point corresponds to the actual starting point of the frame if the combination of the combined comparisons exceeds a predetermined threshold.

21. A method as claimed in any of claims 14 to 20, comprising selecting multiple different potential starting points for the frame, and for each potential starting point, dividing the received signal into consecutive, non-overlapping sections commencing at the potential starting point.

22. A method as claimed in claim 21 , comprising generating a combination of combined comparisons in respect of each starting point and determining the actual starting point of the communication in dependence on those combined comparisons.

23. A method as claimed in claim 22, comprising:

determining an average of the combination of combined comparisons for all of the potential starting points; and

determine that a potential starting point is the actual starting point if the combination of combined comparisons in respect of that potential starting point exceeds the average by a predetermined amount.

24. A receiving device configured to detect the start of a frame in a signal by:

comparing a first section of the signal with a known sequence that indicates the start of the frame to generate a first comparison including a first phase component;

25. A method substantially as herein described with reference to the accompanying drawings.

26. A receiving device substantially as herein described with reference to the accompanying drawings.