WO2015165533A1

WO2015165533A1 - Improved carrier recovery

Info

Publication number: WO2015165533A1
Application number: PCT/EP2014/058879
Authority: WO
Inventors: Björn GÄVERT; Filip HELLGREN
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2014-04-30
Filing date: 2014-04-30
Publication date: 2015-11-05

Abstract

A method in a carrier recovery unit for correcting a phase of a received radio signal. The method comprises the steps of receiving (S1) a radio signal with phase noise and interference, and adjusting (S2) a transfer function of an adaptive filter based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal. The method also comprises the steps of determining (S3) a frequency error in the received radio signal, filtering (S4) the determined frequency error with the adaptive filter, thereby generating a filtered frequency error, and correcting (S6) the phase of the received radio signal based on the filtered frequency error.

Description

IMPROVED CARRIER RECOVERY

TECHNICAL FIELD

The present disclosure relates to wireless communication systems, and in particular to a carrier recovery unit and a method for correcting a phase of a received radio signal, as well as a radio receiver comprising the carrier recovery unit.

BACKGROUND

Modern high frequency communication systems, such as fixed microwave point-to-point radio links, can operate in frequency bands ranging from around 3 GHz up to 80 GHz and above. The high frequency carrier signals used by these systems are often generated using microwave local oscillators, LOs, which normally add a significant amount of phase noise to the modulated signal. Thus, high frequency communication systems often suffer from high levels of phase noise. Furthermore, requirements on communication system throughput are constantly increasing due to, e.g., increased user traffic in communication networks. One way to increase throughput is to use dense modulation formats, such as 1024 quadrature amplitude modulation, OAM, which offer high spectral efficiency. However, more dense modulation formats also lead to an increased sensitivity to phase noise. At least partly due to these reasons, phase noise is a factor which limits the throughput of many communication systems, and consequently there is an ever increasing need to achieve better phase noise performance in order to reach higher throughputs at lower costs.

There are two principal approaches of tackling this phase noise problem; improving the actual microwave LO components such that less phase noise is generated, or improving the resilience and robustness of the communication system to phase noise.

Reducing the phase noise level of an LO can be done in a number of ways, such as decreasing the tuning range of the LO, i.e., the range of frequencies which the LO can be configured to output, or by allowing the LO to consume more power. However, if the tuning range of the LO is decreased, more component variants will be needed to cover a given frequency band, which leads to increased cost. Allowing the LO to consume more power is not a desirable solution either, since a high power consumption can lead to increased cost as well as potential hardware cooling problems. Improving the resilience of the communication system to phase noise can also be done in a number of different ways, such as adding advanced channel codes to the receiver which account for the presence of phase noise during decoding of a received radio signal. However, advanced channel codes, e.g., based on sequence detection via Trellis schemes, often add complexity to the radio receiver. Increased radio receiver complexity is problematic due to, e.g., cost reasons, and also due to that development time and time-to-market tend to increase with receiver complexity.

Another countermeasure against phase noise is to implement digital signal processing, DSP, in the radio receiver to track and compensate for a randomly fluctuating phase. Such processing is commonly referred to as carrier recovery processing or simply carrier recovery. Improving the ability of the communication system to track the phase of a received signal by carrier recovery schemes does not necessarily add significant cost to the radio equipment, since silicon area in ASICs or FPGAs for performing this task is often negligible compared to other receiver functions such as channel filtering, equalization, or channel coding.

However, the use of DSP schemes for tracking phase often results in a large number of tuning parameters which must be manually configured for a given operating scenario in order to optimize the signal processing to the current operating conditions.

It is not desirable to have a signal processing with a large number of tunable system parameters since each of the parameters must be manually set and verified before the system can be shipped to customer. This tends to drive development and verification time and thus delays time to market.

Thus, there is a need for DSP-based carrier recovery methods and carrier recovery units which do not require extensive manual configuration, and which are parameterized by as few manually set system parameters as possible. SUMMARY

An object of the present disclosure is to provide a carrier recovery unit, a radio receiver, a method, and a computer program, which seek to mitigate, alleviate, or eliminate one or more of the above-identified deficiencies in the art and disadvantages singly or in any combination and to provide improved carrier recovery.

This object is obtained by a method in a carrier recovery unit for correcting a phase of a received radio signal. The method comprises the steps of receiving a radio signal with phase noise and interference, and automatically adjusting a transfer function of an adaptive filter based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal such that the adaptive filter suppresses the interference but passes the phase noise. The method also comprises the steps of determining a frequency error in the received radio signal, and filtering the determined frequency error with the adaptive filter, in which the adaptive filter suppresses the interference but passes the phase noise of the determined frequency error, thereby generating a filtered frequency error, as well as correcting the phase of the received radio signal based on the filtered frequency error.

By the present teaching, an improvement in the phase noise performance of, e.g., a radio link receiver, is obtained by improving on the correcting of the phase of a received radio signal, i.e., by performing carrier recovery. Consequently, phase noise performance is increased without having to improve on the phase noise level of any communication system hardware, such as a microwave LO.

Furthermore, by the disclosed features of adjusting and filtering, there is provided herein a method for automatically adapting carrier recovery to, e.g., a given oscillator design, carrier frequency, signalling format, and signal-to-noise ratio, SINR. This is achieved since the transfer function of the filter is adjusted based on the time autocorrelation of the phase noise and the interference in the radio signal. These time autocorrelation functions being indicative of, and at least partly determined by, the given oscillator design, the current operating frequency, the signalling format, and the signal to interference and noise ratio, SINR. The present technique further reduces the need for manually setting tuning parameters by the feature of automatically adjusting the transfer function of an adaptive filter, as opposed to manually setting the transfer function of the adaptive filter. This automatic adjustment of the transfer function of the adaptive filter is, as will become apparent from the below description, facilitated by the features of filtering a determined frequency error, and correcting the phase of the received radio signal based on the filtered frequency error, as opposed to filtering a determined phase error and correcting the phase of the received radio signal based on the filtered phase error.

Consequently, the cost, time-to-market, and power consumption of, e.g., a radio receiver implementing the proposed method is potentially reduced, since the present teaching will not drive cost and power consumption of the microwave LOs, nor will it add to parameterization complexity by using an excessive amount of manually configured tuning parameters.

Furthermore, an operator of, e.g., a fixed point-to-point radio link, will not have to re- parameterize signal processing when changing hardware, e.g., when upgrading to a different oscillator design with different phase noise characteristics.

According to an aspect, the interference in the received radio signal comprises any of thermal noise, interference from radio transmitters, and non-linear distortion.

Thus, the present method can be applied in a variety of operating scenarios, and provides a measure of flexibility when it comes to different types of interferences. A normal operating condition often involves thermal noise such as additive white Gaussian noise. However, a combination of interferences could occur, in which case the present technique will adapt accordingly.

According to some aspects, the step of receiving further comprises receiving a plurality of pilot symbols comprised in the received radio signal at pre-determined time and/or frequency locations in the received radio signal, and the step of determining further comprises determining the frequency error based on received pilot symbols.

Thus, according to some aspects, it is proposed herein a carrier recovery scheme which makes use of pilot symbols to track phase, but which avoids the need for an extensive amount of tuning parameters for optimizing the processing of the pilot symbols to fit a given communication scenario, by use of adaptive signal processing applied to the pilot symbols.

An operator of, e.g., a fixed point-to-point radio link, will not have to re-parameterize this pilot signal processing when changing hardware, e.g., when upgrading to a different oscillator design with different phase noise characteristics.

According to aspects, the step of adjusting further comprises adjusting a filtering bandwidth of the adaptive filter.

The feature of adjusting a filtering bandwidth of the adaptive filter is one example of automatically adjusting the carrier recovery to changing operating conditions without need for manual re-tuning of filter parameters.

For example, if the power of an interference component comprised in the received radio signal changes, the step of adjusting according to aspects comprises adjusting the filtering bandwidth to account for the new thermal noise level.

According to aspects, the step of adjusting further comprises automatically decreasing the filtering bandwidth of the adaptive filter when the interference in the received radio signal increases in power, and automatically increasing the filtering bandwidth of the adaptive filter when the interference in the received radio signal decreases in power.

These features improve the performance of the present method for correcting a phase of a received radio signal in that the interference power level is automatically accounted for in the step of adjusting. If the interference power level increases the interference can be suppressed to a further degree by decreasing the filtering bandwidth. However, during operating conditions when there is no high power interference in the received radio signal, the filtering bandwidth is increased, which leads to a faster, i.e., improved, phase noise tracking.

According to further aspects, the step of adjusting comprises automatically decreasing the filtering bandwidth of the adaptive filter when the time autocorrelation of the phase noise in the received radio signal increases, and automatically increasing the filtering bandwidth of the adaptive filter when the time autocorrelation of the phase noise in the received radio signal decreases. These features improve the performance of the current method for correcting a phase of a received radio signal in that the characteristics of the phase noise process are accounted for. When there is high time autocorrelation in the phase noise process, it is beneficial to decrease filtering bandwidth, and vice versa. Thus, the present method allows for automatically adjusting the transfer function of the adaptive filter to current phase noise conditions.

According to some aspects, the step of adjusting further comprises updating the transfer function of the adaptive filter according to a Least-Mean-Squares, LMS, operation.

The fact that the step of adjusting can be based on an LMS operation is advantageous in that the LMS operation can be implemented with limited complexity. Furthermore, the LMS operation often provides good performance and also provides great flexibility in adjusting to different operating conditions. Further, the operating conditions may not be known beforehand.

Furthermore, the LMS operation provides a straight forward way to adjust the transfer function of the adaptive filter based on the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal.

There is also provided a computer program, comprising computer readable code which, when run on an apparatus, causes the apparatus to perform the method disclosed herein.

The object stated above is also obtained by a carrier recovery unit comprising an input port configured to receive a radio signal with phase noise and interference, a frequency error determining unit arranged to determine a frequency error in the received radio signal, and an adaptive filter configured to have an adjustable transfer function. The adaptive filter is arranged to receive the determined frequency error from the frequency error determining unit and to filter the frequency error to suppress interference in the determined frequency error while passing phase noise in the determined frequency error, thereby generating a filtered frequency error.

The carrier recovery unit further comprises a filter update unit arranged to adjust the transfer function of the adaptive filter, based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal, into an adjusted filter transfer function adapted for suppressing interference in a filtered signal, while passing phase noise in the filtered signal, as well as a second phase correcting unit adapted to receive the radio signal, and to receive the filtered frequency error and to correct the phase of the radio signal based on the filtered frequency error.

The object is furthermore obtained by a radio receiver comprising an antenna, a receiver front end unit, the carrier recovery unit of the present teaching, and a second detector unit.

The computer program, the carrier recovery unit, and the radio receiver display advantages corresponding to the advantages already described in relation to the methods performed in the carrier recovery unit.

BRI EF DESCRIPTION OF THE DRAWINGS

Further objects, features, and advantages of the present disclosure will appear from the following detailed description, wherein some aspects of the disclosure will be described in more detail with reference to the accompanying drawings, in which :

Figure 1 is a block diagram illustrating a radio transmitter and a radio receiver. Figures 2a-2d are flowcharts illustrating embodiments of method steps.

Figure 3 is a block diagram illustrating details of a carrier recovery unit.

Figure 4 is a graph showing phase values as function of time.

Figure 5 is a block diagram illustrating details of a radio receiver.

Figure 6 is a block diagram illustrating a carrier recovery unit.

DETAI LED DESCRIPTION

Aspects of the present disclosure will be described more fully hereinafter with reference to the accompanying drawings. The apparatus, computer program and methods disclosed herein can, however, be realized in many different forms and should not be construed as being limited to the aspects set forth herein. Like numbers in the drawings refer to like elements throughout, except for a prefix digit in the number which represents the drawing page in which the element is to be found. The various example embodiments described herein are described in the general context of method steps or processes, which may be implemented in one aspect by a computer program product, embodied in a computer-readable medium, including computer-executable instructions, such as program code, executed by computers in networked environments. A computer-readable medium may include removable and non-removable storage devices including, but not limited to, Read Only Memory (ROM), Random Access Memory (RAM), compact discs (CDs), digital versatile discs (DVD), etc. Generally, program modules may include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Computer-executable instructions, associated data structures, and program modules represent examples of program code for executing steps of the methods disclosed herein. The particular sequence of such executable instructions or associated data structures represents examples of corresponding acts for implementing the functions described in such steps or processes.

The terminology used herein is for the purpose of describing particular aspects of the disclosure only, and is not intended to limit the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

The present technique achieves an improvement in the phase noise performance of a carrier recovery unit, and thus also an improvement in the phase noise performance of a radio receiver implementing the carrier recovery unit, or, equivalently, performing the method in a carrier recovery unit for correcting a phase of a received radio signal proposed herein.

Herein, when discussing the phase of a received radio signal, it is meant an unwanted phase change or jitter due to imperfections in transmitter and/or receiver microwave LOs, and not phase changes due to phase modulation of the transmitted signal in order to carry information between transmitter and receiver. As an example, suppose the transmitter modulates the transmitted signal to carry information using a given signal constellation, e.g., a 1024- AM constellation. A received signal without changes or distortions in phase, i.e., phase noise, will exhibit the same constellation at the detector unit in the receiver as was generated at the transmitter, whereas a signal that is received with phase error or jitter will exhibit a rotated signal constellation at the detector unit in the receiver compared to the generated signal constellation at the transmitter.

The term phase error will also be used herein where appropriate to describe an unwanted phase change or jitter due to imperfections in transmitter and/or receiver microwave LOs. A key concept of the present teaching is that the signal processing in the carrier recovery unit comprises the generation of a frequency error, as opposed to a phase error. This frequency error is processed by a new type of adaptive filter which automatically adapts to changing system conditions.

Herein, time autocorrelation refers to the time autocorrelation function of a signal. The time autocorrelation is computed as a function of delay. The time autocorrelation at zero delay is defined to be the power of the signal. A high time autocorrelation means that the autocorrelation function declines comparably slow with delay, as opposed to a low time autocorrelation which declines comparably fast with delay. A so-called white signal has a time autocorrelation function which is positive at zero delay, and zero elsewhere. An example of a white signal is additive white Gaussian noise, AWGN.

Since the adaptive filter discussed herein automatically adapts its transfer function to optimize performance for, e.g. a current oscillator design, operating frequency, signalling format, and signal-to-noise ratio, SINR, there is no manual tuning of filter parameters needed. For example, the tap values of a finite impulse response, FIR, filter, is automatically set in order to optimize carrier recovery signal processing to a given set of operating conditions without the need for manual configuration.

This is achieved since the transfer function of the filter is adjusted based on the time autocorrelation of the phase noise and the interference in the radio signal. These time autocorrelation functions being indicative of, and at least partly determined by, the given oscillator design, the current operating frequency, the signalling format, and the signal-to- noise ratio, SINR. The design of the oscillators used for up- and down-conversion will have a significant impact on the time autocorrelation properties of the phase noise, the current operating frequency will also have an impact on the time autocorrelation properties of the phase noise, since, the higher the operating frequency the stronger the phase noise, in general. Furthermore, the signalling format will influence detector error rates and therefore also the quality of the determined frequency error. Also, the SINR will be reflected in the time autocorrelation function of the interference.

In order to first describe a context in which the proposed method is likely to be performed, a system overview of a high frequency communication system will first be given, followed by a detailed description of the proposed method.

Figure 1 shows a block diagram illustrating a radio transmitter 110 and a radio receiver 100. The radio transmitter comprises an oscillator 109, herein referred to as a microwave local oscillator, LO. The transmitter microwave LO 109 is used to up-convert a baseband or an intermediate frequency, IF, signal up to radio frequency. The radio frequency can, as mentioned above, be high, i.e., ranging from around 3 GHz all the way up to 80 GHz and above. Consequently, the transmitted radio signal 108 can be expected to be distorted by phase noise which is added during the process of up-conversion in frequency. The high frequency radio signal 108 is arranged to be transmitted from an antenna 111. This antenna unit can be a directive antenna unit such as a disc antenna, which is common in fixed point-to- point radio links. However, the antenna 111 can also be an omni-directional antenna, or an antenna array of some sort.

The high frequency radio signal 108 is arranged to be received by a radio receiver 100. The radio receiver 100 receives the high frequency radio signal 108 via an antenna 112, which antenna 112 can be a directive antenna unit such as a disc antenna. However, as with the antenna unit of the transmitter 110, the receiver antenna 112 can also be an omnidirectional antenna, or an antenna array of some sort.

The radio signal captured by the antenna 112 is fed to a receiver front-end unit 114. The receiver front-end unit 114 also comprises a microwave LO 113, which microwave LO 113 is used to down-convert the high frequency radio signal to baseband, or to an IF. As for the oscillator 109 at the transmitter 110, the oscillator 113 at the receiver 100 is likely to add an amount of phase noise to the down-converted signal due to the high frequencies involved. It is observed that, in case the microwave LO 113 at the receiver end is used to down-convert to IF, then there will be additional oscillators comprised in the receiver 100 which are used to further down-convert the IF signal to baseband for further processing.

The receiver front-end unit 114 is arranged to output a received radio signal 103, which according to aspects, is a baseband signal. This received radio signal 103 comprises phase noise due to the up- and down-conversion, and it is also likely to comprise interference.

The interference can have originated from various sources. Most likely the interference will comprise thermal noise, such as additive white Gaussian noise, AWGN. This AWGN can be expected to have a low time autocorrelation. The interference can also comprise narrow-band interference from other radio transmitters having higher time autocorrelation, and also nonlinear distortion.

Thus, according to some aspects, the interference in the received radio signal 103 comprises any of thermal noise, interference from radio transmitters, and non-linear distortion.

The radio receiver 100 further comprises a carrier recovery unit 101, which carrier recovery unit is arranged to receive the radio signal 103 from the receiver frond-end 114 on an input port 105, and to output a phase corrected radio signal 104 on an output port 106, which phase corrected radio signal 104 is arranged to be received by a detector unit 115.

There is also shown a feedback signal 122 from the detector unit 115 to the carrier recovery unit 101, which feedback signal will be further elaborated on below. Figure 2a shows a flowchart illustrating embodiments of method steps according to the present technique. In particular, there is illustrated a method in a carrier recovery unit 101 for correcting a phase of a received radio signal 103. The method comprises the step of receiving SI a radio signal 103 with phase noise and interference. This radio signal 103 is the radio signal which is output from the receiver front-end 114, which has undergone up-conversion and down-conversion. This up-conversion and down-conversion is the main source of the comprised phase noise. The interference, as discussed above, can have originated from many different sources.

In order to set up the adaptive filter 102 for filtering, the method comprises the step of automatically adjusting S2 a transfer function of an adaptive filter 102 based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal 103 such that the adaptive filter suppresses the interference but passes the phase noise. Since the transfer function of the adaptive filter is automatically adjusted, there is no need for manual configuration of filter transfer function. Such a manual configuration of adaptive filter transfer function could, e.g., entail a manual setting of the filter taps of a finite impulse response, FIR, filter.

There are various ways of achieving such an automatic adjusting of the transfer function, as will be elaborated on below in connection to Figures 3 and 5. However, in general, the relationship between the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal can be discerned from an error signal generated while attempting to detect information symbols modulated onto the received radio signal.

The time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal can also be discerned from a-priori known pilot symbols embedded in the received radio signal. Such pilot symbol processing would entail comparing phase distortion over time with amplitude distortion over time, thus determining a relative time autocorrelation function between amplitude distortion and phase distortion.

It again noted that the time autocorrelation of a signal includes the power of the signal, since the power of a signal is defined as the time autocorrelation of the signal at zero time offset. Thus, adjusting S2 the transfer function of the adaptive filter 102 based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal comprises adjusting S2 the transfer function of the adaptive filter 102 based on the power of the phase noise and the power of the interference in the received radio signal. According to some aspects, the step of automatically adjusting the transfer function of the adaptive filter 102 may be based on the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal is to implement a least-mean- squares, LMS, operation, or a recursive least squares, RLS, operation based on a detector error signal, as will be further elaborated on below. As already noted above, an alternative to using a detector error signal for automatically adjusting the adaptive filter transfer function, e.g., by LMS or RLS operations, is to measure or otherwise determine the time autocorrelation of the phase noise and the time autocorrelation of the interference by observation of the received radio signal, possibly by observation of a- priori known pilot symbols comprised in the received radio signal.

A further alternative implementation of the step of automatically adjusting the transfer function of the adaptive filter 102 based on the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal is to implement a Wiener filter from a-priori determined time autocorrelations of the phase noise and interference in the received radio signal, which Wiener filter configuration may be a-priori determined from experimentation in lab.

Classical carrier recovery modules often have two consecutive loops to make an iterative carrier recovery scheme; a first carrier recovery loop, FCR, and a second carrier recovery loop, SCR. The phase noise of the received signal can then be measured in the FCR, processed, and then used to compensate the phase in the SCR. This type of system will be further discussed in connection to Figures 3 and 5 below.

A general problem with the approach of using phase measurements from the FCR to compensate phase in the SCR is that the signal which is used to compensate the SCR is a phase signal, and as there are often frequency differences between TX and Rx this signal does not have a well-defined expected value. Even if there are no significant frequency differences between Tx and Rx, phase noise will cause a phase drift which means that the signal which is used to compensate the SCR does not have a well-defined mean value.

This makes it hard to filter the signal and to make the filter adaptive in such a way that it measures the errors and automatically adjusts the filter to current oscillator design, operating frequency, signaling format, and signal-to-noise ratio, SINR.

An alternative, which is proposed here, is to instead measure the frequency error in the FCR, hence a signal with expected value equal to zero can be created. This has two advantages: filtering is easier and it is possible to build an adaptive algorithm which, by adapting the filter parameters, automatically minimizes the degradation due to phase noise. Thus, the method illustrated in Figure 2a further comprises the step of determining S3 a frequency error in the received radio signal 103, and also filtering S4 the determined frequency error with the adaptive filter 102, in which the adaptive filter suppresses the interference but passes the phase noise of the determined frequency error, thereby generating a filtered frequency error.

As noted above, the features of determining a frequency error, instead of a phase error as is more common in carrier recovery methods, is a key concept underlying the present teaching. By determining a frequency error, and filtering this frequency error instead of working with a phase error, the automatic adjustment of the transfer function of the adaptive filter is simplified. This is at least partly because frequency error has a well-defined expected value, which is not true for a phase error that wraps around itself at a value of 360 degrees, or 2 pi in case radians are used. This wrapping property of a phase error complicates automatic adjustment of filter transfer function, which is avoided by the present teaching of using a frequency error. The method further comprises the step of correcting S6 the phase of the received radio signal 103 based on the filtered frequency error. Thus, by the present teaching, an improvement in the phase noise performance of, e.g., a radio link receiver such as the receiver 100 shown in Figure 1, is obtained by improving on the signal processing used for correcting the phase of a received radio signal. Consequently, phase noise performance is increased without having to improve, i.e., reduce, the phase noise level of any communication system hardware, such as a microwave LO 109, 113.

By the present method, there is provided an automatic adaptation of the carrier recovery unit to optimize carrier recovery performance for, e.g., a given oscillator design, operating frequency, signalling format, and signal-to-noise ratio, SINR. Furthermore, no manual tuning of filter parameters is needed in order to optimize signal processing of a received radio signal to a given set of operating conditions.

This is achieved since the transfer function of the filter is adjusted based on the time autocorrelation of the phase noise and the interference in the radio signal. These time autocorrelation functions being indicative of, and at least partly determined by, the given oscillator design, the current operating frequency, the signalling format, and the signal-to- noise ratio, SINR.

The present disclosure reduces the need for manually set tuning parameters by the feature of automatically adjusting the transfer function of an adaptive filter, as opposed to manually setting the transfer function of the adaptive filter. This automatic adjustment of the transfer function of the adaptive filter is facilitated by the disclosed features of filtering a determined frequency error, and correcting the phase of the received radio signal based on the filtered frequency error, as opposed to filtering a determined phase error and correcting the phase of the received radio signal based on the filtered phase error. Consequently, the cost, time-to-market, and power consumption of, e.g., a radio receiver implementing the proposed method is likely to decrease, since the present phase noise suppression method will not drive cost and power of the microwave LOs, nor will it add to parameterization complexity by using an excessive amount of manually configured tuning parameters. The solution proposed herein can furthermore enable higher capacities in a communication system without penalizing, e.g., bit error rate, BER, performance.

The solution will adapt itself to account for current oscillator design, operating frequency, signaling format, and SINR, to name a few. Thereby complexity is reduced when it comes to configuration and verification.

Furthermore, a customer will not have to re-parameterize pilot signal processing when changing hardware, e.g., when upgrading to a better oscillator design.

Turning again to Figure 2a, there are shown certain aspects of the proposed method. According to one such aspect, the method further comprises the step of delaying S5 the received radio signal 103. The step of correcting S6 then comprises correcting the phase of the delayed radio signal based on the filtered frequency error. The technique involving delaying of the received radio signal will be further discussed in connection to Figures 3 and 5 below. However, it is noted that by the features of delaying the received radio signal, and correcting the phase of the delayed radio signal, a further performance improvement is obtained in that the delay in determining the frequency error is compensated for by also delaying the received radio signal before correcting the phase of the received radio signal. Thus, a phase noise disturbance in the received radio signal can be instantaneously corrected in the delayed received radio signal, as opposed to first detecting an error, and not correcting the error until some time has passed. The concept is similar to an iterative receiver, which gradually refines the detection of a received signal in a plurality of consecutive signal processing steps.

According to another such aspect, the method further comprises the step of outputting S7 the phase corrected delayed radio signal as a phase corrected received radio signal 104. This phase corrected delayed radio signal was shown n Figure 1 as the output 104 from the carrier recovery unit 101. As already noted, the phase noise problem is especially relevant in the context of fixed point- to-point radio links, e.g., microwave radio links used for backhaul in cellular networks. Consequently, according to one aspect, the radio signal is a single carrier radio signal, and the step of receiving SI further comprises receiving Sll the single carrier radio signal over a fixed point-to-point radio link. Many carrier recovery schemes make use of pilot symbols, i.e., known or unknown symbols different in some way from the regular data payload symbols, in order to track a randomly fluctuating phase. Pilot symbols are symbols which are embedded into a stream of regular data payload symbols. The pilot symbols are, in general, easier to detect than the regular data payload symbols, either because they are known a-priori, or, e.g., because they are chosen from a less dense symbol constellation.

Pilot symbols can be used in different ways and in different receiver implementations.

For instance, pilot symbols can be unknown but chosen from a less dense symbol constellation than the data payload symbol constellation. As an example of using pilot symbols chosen from a less dense symbol constellation, suppose a received radio signal comprises data symbols modulated using a 1024-QAM constellation, but every 20:th symbol is instead modulated using a 4-QAM constellation. Thus, at a high enough SINR allowing detection of the 1024-QAM modulated signal at some acceptable error rate, it is unlikely that many decision errors occur when detecting the pilot symbols, and therefore the error signal generated from the pilot symbols is more reliable than the error signal which can be generated from detection of the regular data information symbols.

Another example of using pilot symbols is to transmit one or more known symbols from a given constellation. Pilot symbols introduce signaling overhead, but can contribute to improving resilience to phase noise. Pilot symbols can also comprise a pilot tone situated in-band with respect to the received radio signal or out-of-band with respect to the received radio signal.

Thus, according to one aspect, the step of receiving SI further comprises receiving S12 a plurality of pilot symbols comprised in the received radio signal at pre-determined time and/or frequency locations in the received radio signal.

Turning now to Figure 2b, where aspects of the step of adjusting S2 are shown.

The transfer function of the adaptive filter 102 shown in Figure 1 is herein proposed to be adjusted based on the operating conditions of the carrier recovery unit 101. For example, in case there is strong interference, i.e., interference of high power, comprised in the received radio signal, then the determined frequency error signal can be expected to be noisy or otherwise distorted. In this case it is preferred to filter the determined frequency error signal using a filter with comparably narrow bandwidth, in order to suppress as much of the noise as possible. However, a down-side to using a narrow bandwidth filter is that fast fluctuations in frequency error are also filtered out. Hence, a trade-off is necessary between filtering to suppress interference in the frequency error signal, and not filtering too much to suppress fast fluctuations in the frequency error signal.

According to some aspects, the step of adjusting S2 further comprises adjusting S21 a filtering bandwidth of the adaptive filter 102.

Herein, the bandwidth of a filter is defined as the total width of all frequency bands which are allowed to pass the filter without significant attenuation, e.g., 3dB above minimum attenuation. Thus, a filter having narrow filtering bandwidth is configured to pass less frequency components in a filtered band compared to a filter having a wider bandwidth. For an FIR filter, the highest filtering bandwidth is obtained when the filter has one of its taps set to a given value, and all other taps set to zero. A smaller filtering bandwidth is obtained when all taps are set to the same, non-zero value.

According to one such aspect, the step of adjusting S2 further comprises automatically decreasing S21a the filtering bandwidth of the adaptive filter 102 when the interference in the received radio signal increases in power, and automatically increasing S21b the filtering bandwidth of the adaptive filter 102 when the interference in the received radio signal decreases in power.

Thus, the filtering bandwidth is decreased in order to suppress strong interference in order to clean up the determined frequency error from distortion, but only when necessary, since the bandwidth is increased when interference decreases in power. In this way the signal processing is adjusted according to current conditions. The reasons for not always using a narrow filtering bandwidth is that fast fluctuations in the frequency error is lost when filtered with a narrow bandwidth filter, which impairs the ability of the carrier recovery to follow fast changes in phase in the received radio signal. According to one other such aspect, the step of adjusting S2 further comprises automatically decreasing S22a the filtering bandwidth of the adaptive filter 102 when the time autocorrelation of the phase noise in the received radio signal increases, and also automatically increasing S22b the filtering bandwidth of the adaptive filter 102 when the time autocorrelation of the phase noise in the received radio signal decreases. Thus, the filtering bandwidth is decreased in order to make use of any time correlation in the phase noise processes at the transmitter and/or at the receiver oscillators. The rationale for this being that a signal with time correlation tends to exhibit similar values over time. Hence, it makes sense to average the signal in order to refine the signal quality, i.e., using a narrower filtering bandwidth. However, if there is no significant time correlation in the signal to be filtered, then it does not make sense to average, i.e., to use a narrow filtering bandwidth. In this way the bandwidth of the adaptive filter is automatically adjusted to the current operating conditions of the carrier recovery unit. Consequently, by the present method, there is provided an automatic adaptation of the carrier recovery unit to optimize carrier recovery performance for, e.g., a given oscillator design, operating frequency, signalling format, and signal-to-noise ratio, SINR. Furthermore, no manual tuning of filter parameters is needed in order to optimize signal processing of a received radio signal to a given set of operating conditions.

One example of the adjusting of the transfer function of the adaptive filter 102 based on the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal is to implement a least-mean-squares, LMS, operation, or a recursive least squares, RLS, operation based on a detector error signal. Thus, according to some aspects, the step of adjusting S2 further comprises updating S23 the transfer function of the adaptive filter 102 according to a Least-Mean-Squares, LMS, operation, and, according to another aspect, the step of adjusting S2 further comprises updating S24 the transfer function of the adaptive filter 102 according to a Recursive-Least- Squares, RLS, operation. The fact that the step of adjusting S2 can be based on LMS or RLS operations is advantageous in that these operations can be implemented with limited complexity. Furthermore, the LMS operation often provides good performance and also provides flexibility in adjusting to different operating conditions such as varying levels of phase noise. Furthermore, these operating conditions may not be known beforehand. The RLS operation is somewhat similar to the LMS operation, and is associated with the same benefits.

Figure 2c shows further aspects of the step of adjusting S2. In particular, there is shown an aspect wherein the step of adjusting S2 further comprises detecting S25a modulation symbols comprised in the phase corrected received radio signal 104 with a second detector unit 115, and determining S25b a phase for each detected modulation symbol based on a comparison between detected modulation symbol values and detector unit 115 input phase corrected received radio signal samples. The step of adjusting S2 also comprises accumulating S25c the determined phase between two consecutive received pilot symbols, and correlating S25d the accumulated phase against the determined frequency error input to the adaptive filter 102, as well as adjusting 25e the transfer function of the adaptive filter 102 based on the correlation between the accumulated phase and the determined frequency error input to the adaptive filter 102.

The aspect shown in Figure 2c is an embodiment of a system in which updating S23 the transfer function of the adaptive filter 102 is done according to a Least-Mean-Squares, LMS, operation.

Turning now to Figure 2d, where aspects of the step of determining S3 are shown.

According to one such aspect, the step of determining S3 further comprises determining S31 the frequency error based on received pilot symbols. As an example, the frequency error may be determined by first determining the respective phases of the received radio signal at the locations of the received pilot symbols. These phases can be determined, e.g., as a phase correction value applied to the received radio signal, in addition to a phase error which is measured by a detector attempting to detect the pilot symbol, which phase determination process will be further discussed in connection to figures 3 and 5 below. The phases can also be directly determined from observing the received pilot symbols in case the phases of the transmitted pilot symbols are known a-priori.

The determined phases of the received pilot symbols will, in case there is a frequency offset between transmitter and receiver, display a trend, i.e., the phases will not be stationary centered on a fixed phase value. In order to get a phase signal with a near-constant mean value, it is proposed herein to subtract a frequency difference estimate from the determined phases. Thus a phase signal with near-constant mean is obtained. The frequency error signal, with respect to the frequency difference estimate, can then be determined from the difference between consecutive values of the stationary phase signal.

The frequency difference estimate can be obtained from a number of different sources, one example being a low-pass filtered frequency state of a loop filter, as will be further discussed in connection to Figures 3 and 5 below.

Thus, according to some aspects, the step of determining S3 further comprises determining S32a a phase of each of the received pilot symbols, and also estimating S32b a trend in the determined phases of the received pilot symbols, as well as subtracting S32c the determined trend from the determined phases of the received pilot symbols, thereby generating a stationary phase error signal. The step of determining S3 then also comprises determining S32d the frequency error as a difference between consecutive values of the stationary phase error signal. It is noted that the method described herein can be implemented by a carrier recovery unit 102 such as the carrier recovery unit 102 shown in Figure 1, and also by the carrier recovery units shown in Figures 3 and 5 and discussed below. However, the present method can also be implemented by a computer program, comprising computer readable code which, when run on a carrier recovery unit 101, causes the carrier recovery unit 101 to perform the method as claimed herein.

Figure 3 shows a carrier recovery unit 301 comprising an input port 305 configured to receive a radio signal 303 with phase noise and interference as elaborated on above. The received radio signal 303 is branched in two parts or copies marked in Figure 3 as 303' and 303".

According to some aspects, the two parts 303', 303" are identical copies of the radio signal 303.

A first such part 303' is input to a first phase correcting unit 316 which is arranged to perform a coarse first phase correction of the received radio signal in order for a first detector unit 317 to receive the coarsely phase corrected radio signal and make a detection of information symbols comprised in the received radio signal 303'. It is again remarked that, when discussing the phase of a received radio signal, it is meant an unwanted phase change or jitter due to imperfections in transmitter and/or receiver microwave LOs, and not phase changes due to phase modulation of the transmitted signal in order to carry information between transmitter and receiver. A received signal without changes or distortions in phase will have the same constellation at the detector unit in the receiver, whereas a signal that is received with phase error will have a rotated signal constellation at the detector unit in the receiver.

The first detector unit 317 is arranged to determine a phase error Pe[nl], i.e., an estimate of the phase of the received radio signal 303, for each received and detected information symbol by comparing the detected symbol value to the corresponding received radio signal sample. I.e., suppose the nl:th input signal to the detector unit is a [nl]*e^J , and the corresponding detected symbol value is a[nl]'*e^jf[n11 , then the error signal generated by the detector unit 317 is Pe[nl]=f[nl]-f[nl]'. I.e., the difference in phase between received sample and detected value is the determined phase error, which is also an estimate of the phase of the received radio signal 303.

According to some aspects, the modulation symbols further comprise pilot symbols embedded into the received radio signal 303. The applied phase correction value by the first phase correcting unit 316 summed with the corresponding symbol phase error as determined by the first detector unit 317 constitutes the phase of the received information symbol. The carrier recovery unit 301 further comprises a frequency error determining unit 318 arranged to determine a frequency error in the received radio signal based on the applied phase correction value 351 by the first phase correcting unit 316 and the phase error 350 determined by the first detector unit 317.

Thus, the frequency error determining unit 318 is arranged to first determine the phase of the received radio signal. This is achieved as explained above, i.e., by summing the phase correction values applied by the first phase correcting unit 316 to the corresponding phase error values as determined by the first detector unit 317. The determined phase at time instant n will henceforth be referred to as Tp[n], i.e., Tp[n]=Ap[n]+Pe[n], where Ap[n] is the phase correction applied by the first phase correcting unit 316, and Pe[n] is the phase error determined by the first detector unit 317.

It is observed that any delay between phase correction by the first phase correcting unit 316 and detection by the first detector unit 317 shall preferably be compensated for prior to summation, in order to sum a phase correction value Ap[n] with a corresponding phase error value Pe[n]. Now, as visualized in Figure 4, The determined phase values Tp[n] for different n will not be stationary in the sense that the phase values will be centered around a given fixed mean phase value, but the phase values will display a trend which indicates the frequency difference between transmitter and receiver. To get a stationary signal, referred to herein as TpS[n], an estimate of this trend, shown in Figure 4 as Fp[n], is subtracted from Tp[n]. I.e., TpS[n] = Tp[n] - Fp[n]. An example of how to determine Fp[n] will be given in connection to Figure 5 below. By this de-trending operation, a signal with a constant expected value is obtained. The idea is illustrated in Figure 4, where phase is shown on one axis, and time on another axis. An example of determined phase values Tp[n] are plotted, and it is seen that the Tp[n] values exhibit an increasing trend which means that there is a frequency offset between transmitter and receiver oscillators. Furthermore, if there is a drift in Tp[n], there is no well-defined mean value associated with Tp[n]. However, after de-trending by Fp[n] a stationary signal TpS[n] with well-defined mean value is always obtained.

The frequency error determining unit 318 then determines the difference between consecutive values of the TpS[n] signal. This gives a frequency error estimate Fe[n], Fe[d] = TpS[n] - TpS[n-l], which has zero mean and can be conveniently filtered by an adaptive filter.

Turning again to Figure 3, the carrier recovery unit 301 further comprises an adaptive filter 302 configured to have an adjustable transfer function. The adaptive filter 302 is arranged to receive the determined frequency error from the frequency error determining unit 318 and to filter the frequency error to suppress interference in the determined frequency error while passing phase noise in the determined frequency error, thereby generating a filtered frequency error 323. In an example implementation, the adaptive filter 302 is a finite impulse response, FIR, filter with a transfer function determined by a pre-determined number of filter tap values. In this case, adjusting the transfer function of the adaptive filter comprises adjusting the values of the different filter taps.

The bandwidth of such an FIR filter is defined as the total width of all frequency bands which are allowed to pass the filter without significant attenuation, e.g., 3dB above minimum attenuation. Thus, a filter having narrow filtering bandwidth is configured to pass a smaller number of frequency components in a filtered band compared to a filter having a wider bandwidth. For an FIR filter, the highest filtering bandwidth is obtained when the filter has one of its taps set to a given value, and all other taps set to zero. A smaller filtering bandwidth is obtained when all taps are set to the same, non-zero value.

The adaptive filter 302 is arranged to be updated by a filter update unit 319. The filter update unit 319 is arranged to adjust the transfer function of the adaptive filter 302, based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal 303, into an adjusted filter transfer function adapted for suppressing interference in a filtered signal, while passing phase noise in the filtered signal. According to one example, said automatic adjustment is achieved by implementing either of an LMS operation or an RLS operation based on an error signal 322 determined by a detector unit external to the carrier recovery unit, either of which provides for an automatic mechanism for determining the transfer function of the adaptive filter 302.

A second part 303" of the received radio signal 303 is fed into a signal delay unit 320 arranged to delay the received radio signal 303" by a delay corresponding to the total signal processing delay incurred by the first phase correcting unit 316, the first detector unit 317, the frequency error determining unit 318, and the adaptive filter 302.

The carrier recovery unit 301 further comprises a second phase correcting unit 321 adapted to receive the radio signal from the signal delay unit 320, and to also receive the filtered frequency error 323 and to correct the phase of the radio signal based on the filtered frequency error. The phase of the radio signal may be corrected based on the filtered frequency error 323 by the second phase correcting unit 321 in various ways.

According to one example implementation, the second phase correcting unit 321 has an internal frequency state, which frequency state determines how much, and with what sign, the received radio signal is rotated. If the filtered frequency error is positive, the second phase correcting unit 321 will decrease its internal frequency state. If the filtered frequency error is negative, the second phase correcting unit will increase its internal frequency state.

According to another example implementation, the second phase correcting unit 321 implements an independent phase tracking system, and the filtered frequency error 323 is used to correct the applied rotation in phase by the independent phase tracking system. Thus, from looking at Figure 3, it can be seen that the carrier recovery unit 301 according to aspects comprises a first carrier recovery section 324 and a second carrier recovery section 325.

The first carrier recovery section 324 is adapted to process the first part 303' of the received radio signal 303 by a first phase correcting unit 316 and a first detector unit 317.

The frequency error determining unit 318 is here arranged to determine the frequency error in the received radio signal from a phase correction value applied by the first phase correcting unit 316, and a phase error determined by the first detector unit 317, and to pass the frequency error onwards for filtering by the adaptive filter 302. The second carrier recovery section 325 receives the filtered frequency error 323, as well as the second part 303" of the received radio signal 303, and proceeds to perform a refined phase correction of the received radio signal 303 based on the filtered frequency error 323.

Turning now to Figure 5, where details of a radio receiver 500 are shown. The radio receiver 500 comprises an antenna 512, a receiver front end unit 514, the carrier recovery unit according to aspects of the present teaching, and also a second detector unit 515. The frequency error determining unit 518 is here shown as a dashed line box.

Thus, an example embodiment of the carrier recovery unit 101 in Figure 1 is shown in Figure 5. However, the carrier recovery unit of Figure 5 is just one example of how the carrier recovery unit 101 in Figure 1 can be implemented. As in the example of Figure 3, the received radio signal 503 in Figure 5 is split into two parts 503', 503". The first part 503' is fed into a first phase correcting unit 516'.

The first phase correcting unit 516' is arranged to receive a signal, and to adjust the phase of the received signal based on a phase value input signal 531. The phase value input signal 531 is here arranged to be generated by a loop filter 530, which, according to aspects, is a second order loop filter, i.e., the loop filter has a phase state, and also a frequency state. The loop filter 530 is arranged to be driven by a phase error signal 533 generated by a first detector unit 517, as has been discussed previously. The phase value 531, the frequency state value 532, and the phase error 533 are all fed to a phase calculation unit 534. The phase calculation unit 534 is arranged to determine the phase of the received radio signal at time instant n as discussed above, i.e.,

Tp[n]=Ap[n]+Pe[n], where Ap[n] is the phase correction value applied by the first phase correcting unit 516', and Pe[n] is the phase error 533 determined by the first detector unit 517.

Now, as was visualized in Figure 4, the determined phase values Tp[n] for different n will not necessarily be stationary in the sense that the phase values will be centered around a given fixed mean phase value, but the phase values may display a trend which indicates the frequency difference between transmitter and receiver.

To get a stationary signal, referred to herein as TpS[n], an estimate of this trend, shown in Figure 4 as Fp[n], is subtracted from Tp[n] by the phase calculation unit 534. I.e., TpS[n] = Tp[n] - Fp[n].

Here Fp is determined from the frequency state of the loop filter 530, preferable after low- pass filtering. Thus, according to an aspect, Fp[n] is determined by low-pass filtering the frequency state of the loop filter 530.

By this de-trending operation, a signal with a constant expected value is obtained. The idea was illustrated in Figure 4, where phase is shown on one axis, and time on the other axis. An example of determined phase values Tp[n] are plotted, and it is seen that the Tp[n] values exhibit an increasing trend. However, after de-trending by Fp[n] a stationary signal with well- defined mean value is obtained.

The frequency estimator unit 535 is arranged to determine the difference between consecutive values of the TpS[n] signal. This gives a frequency error estimate Fe[n], 537,

Fe[n]=TpS[n] - TpS[n-l], which has zero mean and can be conveniently filtered by the adaptive filter 502.

The frequency error signal 537 is arranged as input to the adaptive filter 502, which is arranged to output a filtered frequency error signal to a second loop filter 536, which second loop filter is arranged to provide a second phase correcting unit 538 with a phase value signal for correcting the phase of the delayed received radio signal, i.e., the output signal from the signal delay unit 520.

According to one example implementation, the second loop filter 536 has an internal frequency state, which frequency state determines how much, and with what sign, the received radio signal is rotated by the second phase correcting unit 538. If the filtered frequency error is positive, the second loop filter 536 will decrease its internal frequency state. If the filtered frequency error is negative, the second loop filter 536 will increase its internal frequency state.

According to another example implementation, the second loop filter 536 also has an internal frequency state, and the filtered frequency error is added as an impulse signal to the frequency state of the second loop filter 536 such that the change is frequency state corresponds to the filtered frequency error.

The radio receiver 500 further comprises a second detector unit 515, which detector unit 515 is arranged to receive the output from the second phase correcting unit 538, i.e., the phase corrected received radio signal 504.

The second detector unit 515 is arranged to detect information symbols comprised in the received radio signal, and to generate an error signal by comparing input samples to the detector unit 515 to detected symbol values. Thus a second phase error signal 522 is generated. This phase error signal is arranged as input to the second loop filter 536, as well as input to the filter update unit 519.

Thus, in other words, the second detector unit 515 is arranged to determine the second phase error signal Pe2[n2] for each received and detected information symbol by comparing the detected symbol value to the corresponding received radio signal sample. I.e., suppose the n2:th input signal to the detector unit is a[n2]*e^jf2[n21, and the corresponding detected symbol value is a[n2]'*e^jf2[n21 , then the second phase error signal generated by the second detector unit 515 is Pe2[n2]=f2[n2]-f2[n2]'. I.e., the difference in phase between received sample and detected value is the determined phase error.

Thus, according to one aspect, the second detector unit 515 is arranged to detect symbols comprised in the phase corrected received radio signal 504, and also to determine a phase error for each detected symbol based on a comparison between detected symbol values and detector unit 515 input phase corrected received radio signal samples.

Said symbols are, according to some aspects payload symbols, and according to some other aspects pilot symbols, and according to some further aspects both payload and pilot symbols. The filter update unit 519 is, according to aspects, arranged to accumulate the determined phase between two consecutive received pilot symbols, and to correlate the accumulated phase against the determined frequency error input to the adaptive filter.

Further, the filter update unit 519 is, according to aspects, arranged to adjust the transfer function of the adaptive filter 502 based on the correlation between the accumulated phase and the determined frequency error input to the adaptive filter 502.

Thus, the filter update unit 519 is arranged to update the transfer function of the adaptive filter 502 according to a Least-Mean-Squares, LMS, operation.

Consequently, a straight forward way of automatically adjusting the transfer function of the adaptive filter 502 based on the time autocorrelation of the phase noise and the time autocorrelation of the interference in the received radio signal is to implement a least-mean- squares, LMS, operation, or a recursive least squares, RLS, operation based on the detector error signal 522.

As an alternative to using the detector error signal 522 for automatically adjusting the adaptive filter transfer function, e.g., by LMS or RLS operations, one can also measure or otherwise determine the time autocorrelation of the phase noise and the time autocorrelation of the interference by observation of the received radio signal, possibly by observation of a- priori known pilot symbols comprised in the received radio signal.

Figure 6 shows a carrier recovery unit S4 for correcting a phase of a received radio signal. The carrier recovery unit comprises: - a first module S41 configured to receive a radio signal 103 with phase noise and interference,

a second module S42 configured to automatically adjust a transfer function of an adaptive filter 102 based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal 103 such that the adaptive filter suppresses the interference but passes the phase noise, a third module S43 configured to determine a frequency error in the received radio signal 103,

a fourth module S44 configured to filter the determined frequency error with the adaptive filter 102, in which the adaptive filter suppresses the interference but passes the phase noise of the determined frequency error, thereby generating a filtered frequency error, and

a fifth module S45 configured to correct the phase of the received radio signal 103 based on the filtered frequency error.

Claims

A method in a carrier recovery unit (101) for correcting a phase of a received radio signal (103), the method comprising the steps of: receiving (SI) a radio signal (103) with phase noise and interference, automatically adjusting (S2) a transfer function of an adaptive filter (102) based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal (103) such that the adaptive filter suppresses the interference but passes the phase noise, determining (S3) a frequency error in the received radio signal (103), filtering (S4) the determined frequency error with the adaptive filter (102), in which the adaptive filter suppresses the interference but passes the phase noise of the determined frequency error, thereby generating a filtered frequency error, and correcting (S6) the phase of the received radio signal (103) based on the filtered frequency error.

The method according to claim 1, further comprising the step of delaying (S5) the received radio signal (103), and wherein the step of correcting (S6) comprises correcting the phase of the delayed radio signal based on the filtered frequency error.

The method according to claim 2, further comprising the step of outputting (S7) the phase corrected delayed radio signal as a phase corrected received radio signal (104).

The method according to any preceding claim, wherein the radio signal is a single carrier radio signal, the step of receiving (SI) further comprising receiving (Sll) the single carrier radio signal over a fixed point-to-point radio link.

The method according to any preceding claim, the step of receiving (SI) further comprising receiving (S12) a plurality of pilot symbols comprised in the received radio signal at pre-determined time and/or frequency locations in the received radio signal. The method according to claim 5, wherein the step of determining (S3) further comprises determining (S31) the frequency error based on received pilot symbols.

The method according to any of claims 5-6, wherein the step of determining (S3) further comprises: determining (S32a) a phase of each of the received pilot symbols, estimating (S32b) a trend in the determined phases of the received pilot symbols, subtracting (S32c) the determined trend from the determined phases of the received pilot symbols, thereby generating a stationary phase error signal, and determining (S32d) the frequency error as a difference between consecutive values of the stationary phase error signal.

The method according to any preceding claim, wherein the step of adjusting (S2) further comprises adjusting (S21) a filtering bandwidth of the adaptive filter (102).

The method according to claim 8, the step of adjusting (S2) further comprising: automatically decreasing (S21a) the filtering bandwidth of the adaptive filter (102) when the interference in the received radio signal increases in power, and automatically increasing (S21b) the filtering bandwidth of the adaptive filter (102) when the interference in the received radio signal decreases in power.

The method according to any of claims 8-9, the step of adjusting (S2) further comprising: automatically decreasing (S22a) the filtering bandwidth of the adaptive filter (102) when the time autocorrelation of the phase noise in the received radio signal increases, and automatically increasing (S22b) the filtering bandwidth of the adaptive filter (102) when the time autocorrelation of the phase noise in the received radio signal decreases.

11. The method according to any of the preceding claims, wherein the step of adjusting (S2) further comprises updating (S23) the transfer function of the adaptive filter (102) according to a Least-Mean-Squares, LMS, operation.

12. The method according to any of claims 1-10, wherein the step of adjusting (S2) further comprises updating (S24) the transfer function of the adaptive filter (102) according to a

Recursive-Least-Squares, RLS, operation.

13. The method according to any preceding claim, wherein the interference in the received radio signal (103) comprises any of thermal noise, interference from radio transmitters, and non-linear distortion.

14. A computer program, comprising computer readable code which, when run on a carrier recovery unit (101), causes the carrier recovery unit (101) to perform the method as claimed in any of claims 1-13.

15. A carrier recovery unit (301) comprising: an input port (305) configured to receive a radio signal (303) with phase noise and interference, a frequency error determining unit (318) arranged to determine a frequency error in the received radio signal, an adaptive filter (302) with an adjustable transfer function, the adaptive filter (302) being arranged to receive the determined frequency error from the frequency error determining unit (318) and to filter the frequency error to suppress interference in the determined frequency error while passing phase noise in the determined frequency error, thereby generating a filtered frequency error (323), a filter update unit (319) arranged to adjust the transfer function of the adaptive filter (302), based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal (303), into an adjusted filter transfer function adapted for suppressing interference in a filtered signal, while passing phase noise in the filtered signal, a second phase correcting unit (321) adapted to receive the radio signal, and to receive the filtered frequency error (323) and to correct the phase of the radio signal based on the filtered frequency error.

The carrier recovery unit (301) according to claim 15, further comprising a signal delay unit (320) arranged to delay the received radio signal (103), wherein the second phase correcting unit (321) is adapted to receive a delayed radio signal from the signal delay unit (320), and to receive the filtered frequency error (323) and to correct the phase of the delayed radio signal based on the filtered frequency error.

The carrier recovery unit (301) according to claim 15 or 16, comprising a first carrier recovery section (324) adapted to process the received radio signal (303) by a first phase correcting unit (316) and a first detector unit (317), the frequency error determining unit (318) being arranged to determine the frequency error in the received radio signal from a phase correction value applied by the first phase correcting unit (316), and a phase error determined by the first detector unit (317).

A radio receiver (100) comprising an antenna (112), a receiver front end unit (114), the carrier recovery unit (301) of any of claims 15-17, and a second detector unit (115).

The radio receiver (100) according to claim 18, the second detector unit (115) being arranged to detect modulation symbols comprised in the phase corrected received radio signal (103), and also to determine a phase error (122) for each detected modulation symbol based on a comparison between detected modulation symbol values and detector unit (115) input phase corrected received radio signal samples, the filter update unit (319) being arranged to accumulate the determined phase between two consecutive received pilot symbols, and to correlate the accumulated phase against the determined frequency error input to the adaptive filter, the filter update unit (319) being arranged to adjust the transfer function of the adaptive filter (302) based on the correlation between the accumulated phase and the determined frequency error input to the adaptive filter (102). r recovery unit (S4) for correcting a phase of a received radio signal, comprising: a first module (S41) configured to receive a radio signal (103) with phase noise and interference, a second module (S42) configured to automatically adjust a transfer function of an adaptive filter (102) based on a time autocorrelation of the phase noise and a time autocorrelation of the interference in the received radio signal (103) such that the adaptive filter suppresses the interference but passes the phase noise, a third module (S43) configured to determine a frequency error in the received radio signal (103), a fourth module (S44) configured to filter the determined frequency error with the adaptive filter (102), in which the adaptive filter suppresses the interference but passes the phase noise of the determined frequency error, thereby generating a filtered frequency error, and a fifth module (S45) configured to correct the phase of the received radio signal (103) based on the filtered frequency error.