Doppler spread estimation for ofdm systems

[0001] The present invention relates to a digital communication system that uses orthogonal frequency division multiplexing (OFDM), and more specifically relates to the evaluation of frequency shifts in a given OFDM channel due to the Doppler effect.

[0002] In wireless communication, each channel typically exhibits high parameter variability. This may be due to the movement of the transmitter, the movement of the receiver and / or changes in the surroundings of the communication route. For cellular systems such as GSM and WCDMA and for radio systems such as DAB Digital Radio and DVB Digital Television, the main reason for the high dispersion caused by the Doppler effect is the relatively high speed of communication terminals. In DAB and DVB standards, when transmitting information from the transmitter to the receiver, OFDM is required.

[0003] In an OFDM system, each channel contains a number of subcarrier frequencies (hereinafter simply referred to as "carriers"), which are modulated independently, each by separate data. Modulation can be performed according to one of many well-known techniques, it can be Quadrature Amplitude Modulation (QAM) or n- Phase Keying Modulation (n-PSK). The OFDM baseband signal is therefore the sum of these modulated subcarrier frequencies. The baseband signal is in turn used to modulate the main radio frequency (RF) signal. An important aspect of the demodulation of such a signal (i.e. recovery of its signal from the baseband) is processing it using Fast Fourier Transform (FFT). The benefit of communication with the use of OFDM is the ability to send data through channels with high time dispersion, while maintaining a rational level of complication on the receiver side.

[0004] Whether a channel should be considered as having a high dispersion time or not depends on the symbol rate at which the system is operated. As a rule, a channel can be treated as non-dispersive if the effective value (rms) of delay spread in a given channel is less than 10% of the symbol duration. Thus, the benefits of the OFDM application become more noticeable as the data transmission speed with which the system is operated increases, exactly like most emerging systems.

[0005] A way to counter the effects of large delay dispersion in OFDM based systems is to use a guard interval (Gl). Gl (also called "cyclic prefix" or "CP" in the literature) is simply a copy of the last part of the OFDM symbol sent before the current symbol. This is schematically illustrated in Figure 1, showing a number of symbols. One of the exemplary symbols 101 includes the last part 103, transmitted as the part preceding the guard compartment 105 (in the figure time flows from left to right). Similarly, other protective compartments are created, consisting of the terminal parts of the symbols immediately following them.

[0006] It is well-known that in an OFDM-based system, the effect of channel time diffusion can also be avoided, also called inter-symbol interference (ISI), by adjusting the length Gl, T has a length at least equal to (maximum) duration of the impulse response

G channel, marked as T Due to the ability of the OFDM system to counteract the effects of m.

delays with a large spread, it is very useful in so-called Single Frequency Networks (SFN) which can be used in radio broadcasting. (In a single frequency network, transmitters located at certain geographical distances operate at the same frequency. To reduce mutual interference, they are synchronized in time).

[0007] Suppose that the part of the OFDM signal that directly carries the information begins at t = 0, and that the length of the guard period is T if the maximum delay spread in this

G channel is T, the requirement for the start of the Fast Fourier Transform FFT window is m, determined by:

-T + T <tso. (1)

G m [0008] Therefore, where Τ <T it is possible to avoid ISI as long as t is chosen according to the expression m G (1). However, if Τ> T the problem is to choose t such that the ISI impact is minimized. m G

For systems designed for use in an SNF network, the protection period is usually so large that the first of these situations is more likely.

[0009] At present, as explained above, free ISI adoption is possible when Τ <T However, m G requires an exact determination of the starting moment for the part of the signal that carries the information.

For this reason, OFDM receivers include means for estimating the time dependence and for estimating the frequency of the received signal.

[0010] To further improve performance, OFDM system receivers typically include channel estimators whose function is to dynamically determine the channel response. This information is later used by the receiver to create the ability to process the received signal in a way that compensates for the effects of time dispersion occurring on a given channel.

[0011] A conventional way to determine the channel response in an OFDM receiver is to select certain carrier frequencies for use during pilot transmission. Pilot signals contain known information that allows the channel estimator to determine the channel response when operating on the selected carrier frequency by comparing the currently received signal with the signal to be transmitted (i.e. the signal that would be expected to be received under ideal channel conditions). Carrier frequencies on which pilot signals are transmitted are spaced at certain intervals, the size of which allows proper estimation of the channel response for carrier frequencies lying between carrier frequencies with pilot signals, by extrapolation of channel responses specified for carrier frequencies with pilot signals.

[0012] Figure 2 is a block diagram of an example OFDM receiver. An analog signal r (t), generated by receiving and converting downwards (either to an intermediate frequency or to the baseband) of the radio frequency signal is supplied to the analog-to-digital (A / D) converter 201. The digitized signal r (k) is then supplied to the time and frequency relationship estimation unit 203, which generates a preliminary estimate of the time dependencies and frequency shifts of the received signals. (The frequency shift is the difference between the frequency of the transmitted signal and the frequency of the received signal.) This information is fed to the frequency correction unit 205 as well as to the unit 207 removing the protective compartments Gl. The guard interval removing unit 207 also receives output from the frequency correcting unit 205. Based on the best available information on time and frequency dependencies, the unit 207 removing the guard intervals Gl removes the guard intervals Gl and provides the part of the received signal that transfers the information to the unit 209 implementing the FFT Fourier Transform, whose output is delivered to the rest of the receiver, including unit 211, accurately estimating time and frequency relationships, which is able to produce more accurate estimates of time and frequency relationships based on the FFT output. Similarly, more accurate time dependency information is sent back to the Gl removal unit 207 to improve its operation.

[0013] The FFT output from unit 209 is also provided to the channel estimator 213, which by interpolation creates a final estimate of the channel response, as explained above.

[0014] The rate of change of channel parameters is often measured by the amount of scatter resulting from the so-called Doppler effect, or by the maximum Doppler frequency, f D.

Doppler frequency is defined as:

<img file="PL1875696T3_D0001.tif" />

where v is the velocity of the receiver expressed in m / s, fc is the carrier frequency expressed in Hz, and c is the speed of light in vacuum (i.e. approximately 3x10⁸ m / s).

[0015] Here, in order to ensure clarity of discussions, it is important to determine the number of issues associated with the word "Doppler" that appear in contexts related to communication. They are as follows:

1. Pure Doppler shift. This phenomenon can be encountered when dealing with a homogeneous channel, its result is only a frequency error. The error caused by the Doppler frequency shift cannot be distinguished from the error resulting from the fact that the transmitter and receiver use not exactly the same frequencies. Pure Doppler shift is relatively easy to estimate and quite simple to remove. This is effectively implemented by multiplying the received signal by a uniform signal with a frequency changed by the value of the estimated Doppler shift.

2. Doppler scattering. When the communication channel is characterized by multipath propagation, individual waves will reach the receiver at different angles and thus will have different Doppler shifts. The maximum Doppler shift is obtained when the incidence angle is zero or equal to the number Pl (but has different signs for both of these angles) and it is possible to obtain all intermediate Doppler frequencies. Unlike the pure Doppler shift case, the Doppler spread cannot be easily compensated by multiplying by a uniform signal. In communication systems, Doppler spread is often treated as a frequency error that simply cannot be removed, this is considered, for example, when estimating channel frequency. The Doppler effect is therefore very similar to the lack of compensation for errors due to frequency changes.

3. Interference between carrier frequencies (ICI) resulting from Doppler scattering. In OFDM systems, pure Doppler shift is not usually a problem, for the same reason that it is not a problem in single carrier wave systems - it can be easily removed. However, if it is not removed, it will cause the same problem as in single-carrier systems and in addition will cause a shift of the FFT transform spectrum. In a similar way to what happened in single carrier systems, the effects of Doppler scatter cannot easily be countered by using simple multiplication. Instead of the described actions, ICI removal by a comprehensive operation performed in an OFDM receiver is used, after calculating the FFT (simple Doppler shift elimination could rather be done before calculating the FFT) and requires accurate determination of the channel properties.

[0016] The following discussion and descriptions of exemplary embodiments are focused on what is described above in paragraph "2", namely on the amount of Doppler scatter in the OFDM communication system. It is assumed that the frequency shift is removed before calculating the FFT.

[0017] As discussed above, the high degree of Doppler scatter means that the channel properties change rapidly. This in turn means that signal reception becomes more difficult if for proper signal demodulation you need to know specific channel properties (such as phase and amplitude properties). Since it is often possible to determine the highest Doppler frequency value to be reckoned with under typical operating conditions, it is possible to design a receiver based on this Doppler frequency. However, if the actual Doppler frequency value is significantly lower, the design taking into account the worst conditions means that too complex algorithms will be used to estimate channel properties.

[0018] Knowledge of the Doppler frequency can also be used to evaluate how to activate certain algorithms in the receiver. For example, knowledge of Doppler frequencies can be used to:

• determining the path finder method in OFDM receivers, that is, to determine how often it is necessary to search for a new path in an impulse response;

• determining the rate of initiation of algorithms associated with redirections between cells during communication in a cellular network (the higher the Doppler frequency value, the more frequent the use of these algorithms should be, since high Doppler frequency means faster receiver movement); and.

• on single-frequency networks, to determine the pace of scanning for a better frequency to which the current connection could be transferred. Thus, although it is possible to design the receiver taking into account the worst possible doppler scatter, it is usually a rather wasteful approach.

[0019] In addition, in cases where the estimation of channel properties is based on Wiener filters to improve system performance as described in US Patent Application No. 10 / 920,928 entitled "Estimating channel properties using adaptive time estimation" by L. Wilhelmsson and others, you need to know the current Doppler frequency to calculate the properties of the Wiener Filter. Also, even if the estimation of channel properties does not refer to a Wener filter approach, but for example uses filters of varying complexity depending on the degree of difficulty in estimating the channel properties, an estimation of the Doppler frequency is required.

[0020] DAB and DVB are two of many systems in which the rate of change of channel properties can be important. In particular, in the newly developed DVB standards for hand-held devices (DVB-H), it can be expected that the services are carried out when the user stands still, which means that the conditions are similar to those prevailing in a given stationary channel, as well as when the user travels fast vehicle (e.g. car), which means that strong Doppler effects occur. DVB-H is based on OFDM, and one of the fragments of the DVB-H receiver requiring the highest computing power are blocks in which channel properties are estimated. As described in the above-mentioned US Patent Application No. 10 / 920,928, knowledge of Doppler spread can be used to find the right interpolation filter used to estimate channel properties.

[0021] The accuracy of the Doppler estimation will usually improve with the duration of the measurements. However, extending the measurement duration means not only that the current estimate will take longer, but also that the receiver will be less sensitive to rapid changes in the current Doppler spread value. Thus, the time for Doppler estimation should be as short as possible, but still long enough to guarantee the required accuracy.

[0022] Consequently, there is a need for effective estimation of Doppler spread.

[0023] The document "Estimation of speed in communication systems based on OFDM", Schoben H and A, VTC 2002, 22.09.2002, XP 010608905, presents a method of estimating Doppler spread in a given OFDM channel, in which the results of estimates performed on individual frequencies are combined subcarriers.

SUMMARY [0024] It should be emphasized that the terms "comprises" and "containing", as used in this specification, are used to determine the presence of specific features, constant values, steps or components or groups thereof.

In accordance with one aspect of the present invention, the objectives set out above have been achieved, as well as other objectives, using methods and devices that estimate the value of channel Doppler spread in an orthogonal frequency division multiplexing (OFDM) system, the channel having multiple frequencies supporting. The estimation involves selecting a set of two or more carrier frequencies from among multiple carrier frequencies. The Doppler spread value for each of the selected carrier frequencies is estimated. The estimation of the Doppler spread value for the entire channel is created by combining the estimated Doppler spread values ​​for each of the selected carrier frequencies. For example, the Doppler spread value in a given channel can be estimated by averaging the estimated Doppler spread value for each of the selected carrier frequencies.

[0026] In some embodiments, selecting a set of two or more carrier frequencies from a plurality of carrier frequencies boils down to selecting a set of two or more carrier frequencies only from among the frequencies belonging to the set of fixed pilot frequencies associated with that channel.

[0027] The number of carrier frequencies selected may be constant. Alternatively, it can also be a variable A / number, where N is a function of the value representing the spread of delays on a given channel.

[0028] In another aspect, the estimation of the Doppler spread value for each of the selected carrier frequencies is reduced to measurements carried out on the selected carrier frequencies in the time interval T, wherein the time interval T is a function of the meas meas value representing the delay spread in a given channel. In some, but not necessarily all of the embodiments, the number of selected carrier frequencies may be a variable number N, where / V is a function of the value representing the delay spread in a given channel.

BRIEF DESCRIPTION OF THE DRAWINGS [0029] The objectives and advantages of the invention will become apparent after reading the following detailed description, combined with the drawings, in which:

Figure 1 1 schematically shows symbols separated by guard periods in an orthogonal frequency division multiplexing system (OFDM).

Drawing 2 is a block diagram of an example OFDM receiver.

Figure 3 is a signal time chart showing the frame structure according to the DVB-T standard.

Figure 4 is a graph of an example OFDM channel amplitude function for T = 1ps.

m

Drawing 5 is a graph of an example OFDM channel amplitude function for T = 5ps.

m

Figure 6 is a graph of an example OFDM channel amplitude function for T = 10ps.

m

Fig. 7 is a graph of an example OFDM channel amplitude function for T = 20ps.

m

Figure 8 is a schematic of the general procedure for determining Doppler spread in a given OFDM channel.

DETAILED DESCRIPTION [0030] In the following, various features of the invention will be described with reference to the drawings, in which similar elements are designated by the same reference numerals.

[0031] Various aspects of the invention will be described in detail below in connection with a number of exemplary embodiments. To facilitate understanding of the invention, many aspects are described in terms of the sequence of actions to be carried out by means of computer system components. It will be explained that in each of the exemplary embodiments, various actions may be performed by specialized circuits (e.g. appropriately connected discrete logic gates, designed for the implementation of specialized functions), by means of the program whose instructions are executed by one processor or by more processors, or by a combination of both of these methods. Furthermore, the invention may additionally be considered as the content of any computer-readable storage medium, such as permanent memory, magnetic disk, optical disk, or medium utilizing a telecommunications connection (such as a radio link, acoustic link, or optical link) containing the appropriate set of instructions computer that can cause the processor to implement the techniques described here. Thus, various aspects of the invention can be embodied in many different forms and all such forms are analyzed within the scope of the invention. For each of the various aspects of the invention, any of such embodiments may be referred to as "logically configured to" perform the described actions, or alternatively may be "logic" performing specific actions.

[0032] To illustrate various aspects of the invention, a number of embodiments have been described. In one of these aspects, the Doppler frequency reported in the OFDM receiver is estimated. Taking advantage of the fact that the correlation between two frequencies for high selectivity channels will be relatively small compared to channels with flat characteristics, a more accurate estimation of the Doppler effect is presented, in which the Doppler frequency is estimated for several carrier frequencies. In another aspect, due to the fact that it is desirable that the time used for the estimation does not have to be too long, a dynamic Doppler estimation technique is used, which includes adjusting the time allocated to estimate the Doppler effect, carried out on the basis of the estimated delay spread in this channel .

[0033] Thus, in some of the exemplary embodiments described below, the Doppler spread estimation is performed in parallel on several OFDM carrier frequencies. Under certain conditions, this makes it possible to estimate the Doppler effect almost immediately. In some example embodiments, the knowledge of delay spread is used to determine the length of time that must be used to estimate the Doppler effect. And, in some exemplary embodiments, the appropriate number of carrier frequencies to estimate the Doppler effect is determined based on the knowledge of the delay spread in a given channel, thus allowing to approach the optimal operation of the system, using calculations with the least possible complexity. These and other aspects of the invention will be further described in greater detail.

[0034] In order to provide an easier understanding of various aspects of the invention, the following description is based on examples to which data was taken from the digital terrestrial television (DVB-T) standard, (see ETSI EN 300 744 V. 1.4.1 (2001- 01), "Digital Terrestrial Television (DVB); Frame structures, channel coding and modulation for digital terrestrial television, hence the reference to the" ETSI document ". More specifically, it is assumed that the duration of the information-transferring part of the OFDM symbol is T = 896 ps, and that the length of the guard period (Gl) is U

7 "/ 4 = 224 ps. For experts in this field of knowledge it should be clear that these and other specific

The values ​​used in the specification are cited only to better explain the various aspects of the invention, and are not critical at all to the implementation of the invention.

[0035] Currently, in a first aspect of the invention, Doppler spread in a given channel is estimated using continuous pilot signals that are transmitted on a given OFDM channel. For example, in 8k mode, as described in the ETSI document, they are contained in data transmitted on 177 carrier frequencies, forming an OFDM channel. A time relationship diagram showing the frame structure in compliance with DVB-T standards is shown in FIG. 3. The frequency domain extends along the horizontal axis, and the time domain extends along the vertical axis. Each square represents the transmitted data; the vertical position of the square means its transmission time, and the horizontal position of the square means the carrier frequency at which it is transmitted. Blacked squares indicate pilot data. The chart shows how the pilot data is distributed, both in time and frequency domain. In particular, it can be seen that on some carrier frequencies pilot data are never transmitted, on some carrier frequencies pilot data is only transferred periodically, and on other carrier frequencies (e.g., forming columns of black squares), the pilot data is transferred during each transmission . The latter are frequencies carrying pilot data on a continuous basis.

[0036] Doppler spread in a given OFDM channel can be estimated using any conventional, continuous pilot signal using a method that can be used in a single carrier frequency system. A general view of such algorithms can be found, for example, in the publication C. Trepedelenlio et al., Estimating Doppler Spread and Signal Strength in Mobile Communication in Application for Adaptive Transmission and Transmission, "Wirel. Commun. and Mob. Comput., Pp. 221-242, Vol. 1, 2001. Briefly, algorithms can be divided into three categories as follows:

1. Direct methods where the channel response is used directly to assess Doppler spread.

2. Methods using the correlation model as a function of Doppler frequency. By estimating the correlation function, the Doppler spread is estimated.

3. Methods based on Doppler spectrum.

[0037] The techniques disclosed herein find use in any of the above categories. However, in further discussion, attention is focused on direct methods because they are characterized by low complexity. Persons with basic proficiency in this field of knowledge will easily adapt the principles explained in further discussion to other Doppler scattering estimation techniques.

[0038] Direct methods are usually based on consideration of either level crossing speed or zero crossing speed. In the case of testing the speed of exceeding the level, the absolute value of the channel response is used and the exceedances of the specified level (depending e.g. on the average power) are counted. In the case of testing the zero overshoot speed, the actual aibo or imaginary portion of the channel response is used, and zero overshoots are counted either by the real or by the imaginary portion of the channel response. In this sense, methods based on zero crossing testing are more advantageous than methods based on crossing a specified level because they do not require testing the strength of the received signal. Therefore, the following description focuses on these methods, although the idea underlying the techniques disclosed here is also valid in the cases of methods based on the examination of exceedances of a certain level.

[0039] In case of considering the real axis or the imaginary axis exceedances, the expected value of the number of exceedances of the zero value E [N 7 is determined by the equation Eq. (3):

zc = (3) where T is the time at which zero crossings are counted. If both exceedances of the real axis and the imaginary axis are calculated, the expected number of exceedances will be doubled. Depending on whether one axis or two axes are used to estimate the Doppler effect, it is known that in the case of a low Doppler spread value and a relatively low signal-to-noise ratio (SNR), the estimate tends to be overstated. This is because if the true channel response is close to one of the axes, the noise may cause several passes through that axis. Although this is a general problem, it becomes more apparent when the Doppler spread is low (because it means that the true channel response can be close to the axis for a relatively long time) and when the SNR is low (which means an increase in probability causing additional exceedances due to the presence of noise). In order to counteract this disadvantage, it is possible to introduce some kind of hysteresis. For example, the hysteresis can be introduced in such a way that subsequent exceedances of the same axis will not be counted. In addition to the exceptional ease of implementation of this method, it has been experimentally found that if the hysteresis is used, the expected value of the number of exceedances of zero value E [Λ /] is determined by the equation

zc, hyst

E [W] = 2f T (4) zc, hyst D meas [0040] In the case of the equation Esq. (3) or (4) used to estimate the Doppler dispersion, it is easy to see that for a certain number of passes, which is required for recognition estimates that are reliable enough, the measurement time must increase when the Doppler spread is small. Alternatively, if the measurement time is kept constant, the accuracy of the Doppler estimation will increase for larger Doppler spread values.

[0041] Tables 1 and 2 present examples of Doppler frequency estimates for an OFDM system in which each of the estimates results from measurements made only on one of many continuous pilot carrier frequencies. The data contained in Table 1 allow a comparison of the results obtained for three different algorithms used to assess Doppler scatter at 10 dB SNR.

Current f =

h

10Hz

20Hz

50hz

100Hz

<td>Way</td><td></td><td></td><td></td><td></td><td></td>

<td>Only real</td><td> 69 (63)</td><td> 60 (50)</td><td> 74 (38)</td><td> 86 (25)</td><td> 119(21)</td>

<td>Real and imaginary</td><td> 79 (50)</td><td> 67 (45)</td><td> 71 (23)</td><td> 83(18)</td><td> 119(26)</td>

<td>Hysteresis</td><td> 23 (26)</td><td> 27 (23)</td><td> 32(10)</td><td> 60(13)</td><td> 103 (21)</td>

Table 1: Comparison of mean deviation and standard deviation for estimating the Doppler spread value under various current Doppler spread conditions, with a measurement time of 100 ms and SNR = 10 dB [0042] In the first algorithm, only the imaginary axis transitions are used ie only the real part of the signal is used). In the second algorithm, both axis real and imaginary crossings are used. Finally, in the third algorithm, both real and imaginary crossings are exceeded, but the hysteresis described above is also implemented. As can be clearly seen in all cases, the noise causes the Doppler scatter to be overestimated in all cases, but as expected, the largest overestimation occurs in cases of small Doppler scatter values. It can also be seen that the introduction of hysteresis significantly improves both the expected Doppler spread estimation value and its standard deviation.

[0043] In table 2, the corresponding results are shown for the case SNR = 30 dB.

<td>Current f = D</td><td>5 Hz</td><td>10Hz</td><td>20Hz</td><td>50hz</td><td>100Hz</td>

<td>Way</td><td></td><td></td><td></td><td></td><td></td>

<td>Only real</td><td> 12 (15)</td><td> 13(11)</td><td> 24(10)</td><td> 53(12)</td><td> 100(19)</td>

<td>Real and imaginary</td><td> 8.9 (8.6)</td><td> 13(9)</td><td> 23 (9)</td><td> 53(10)</td><td> 100(13)</td>

<td>Hysteresis</td><td> 7.2 (3.6)</td><td> 13(5)</td><td> 22 (8)</td><td> 53(15)</td><td> 100(19)</td>

Table 2: Comparison of the mean deviation and standard deviation for the Doppler spread estimation under various current Doppler spread conditions, with a measurement time of 100 ms and SNR = 30 dB [0044] As expected, the accuracy of the estimation improves, particularly in the estimation expected value. The above-mentioned improvement in estimation accuracy is also noticeable for those cases in which the Doppler spread is high. Taking as an example the algorithm in which the hysteresis was used, the standard deviation from the expected value is reduced from 0.5 to below 0.2 for frequencies in the range of 5

Hz to 100 Hz. If the overshoots are treated as independent, it can be expected that the standard deviation will decrease in the ratio 1 /&gt; lf suggesting that a slightly greater accuracy accuracy may be expected.

[0045] It is clear from Tables 1 and 2 that the Doppler spread expected value is too high in the case where the current Doppler spread is low. This is especially true for low SNR values. It is therefore possible to counteract this estimation disadvantage by slightly adjusting the estimated Doppler spread value by simply using a value slightly lower than the estimated one.

[0046] Tables 1 and 2 show that the suggested estimation method using hysteresis is better than both other methods. This is true when we consider the average value of the estimation, however more it concerns the value of the standard deviation. The deviation of the mean value can be somewhat corrected, as described above, but in this way the increase in variance cannot be countered. Instead, variance must typically be reduced by simply extending the measurement time.

[0047] In an OFDM system in which the channel contains a number of carrier frequencies, it is possible to obtain a more reliable estimation of the Doppler spread for a given channel by estimating the Doppler spread for several different carrier frequencies and then combining the results. For example, a combination of results can be performed by averaging the estimated Doppler spread values ​​determined for individual carrier frequencies. Thus, below the results for the algorithm using hysteresis will be considered, in which all continuous pilot signals are used and these results will be compared with the case when the Doppler effect is estimated using only one carrier frequency.

[0048] To illustrate this aspect of the invention, the 8k mode of the DVB-T standard will again be considered as described in the ETSI document. As previously mentioned, in 17k mode it is possible to provide 177 continuous pilot signals. It is clear that, by estimating the Doppler effect in parallel, on several carrier frequencies, we will always get a positive effect because the effect of noise will be averaged.

Therefore, it can also be expected that if the channel delay spread is small, the accuracy of the Doppler effect estimation will be significantly improved. However, the improvement in accuracy may be even greater if the channel variances for the different carrier frequencies are not substantially correlated with each other (i.e. we will approach this state, or the correlation will be only insignificant).

To realize the tero how big the delay spread should be to get no correlation on the carrier frequencies on which the continuous pilot signals are located, examples of the presumed appearance of channel amplitude functions will be considered when the delay is T = 1ps, 5ps, 10ps and 20ps, as shown in Figs. 4-7, respectively. In each of these drawings, the location of continuous pilot signals is marked with an asterisk.

[0049] FIG 4 is a graph of an exemplary OFDM channel amplitude function dia T = 1ps. Regarding

FIG 4, showing the amplitude functions for T = 1ps, it is easy to see that there is a very strong m correlation between neighboring pilot signals, however, even when T is increased m only to 5ps (FIG. 5), the correlation between neighboring pilot signals is significantly reduced.

In fact, comparing FIG. 5 of FIG. 6 (Γ = 10ps and 20ps, respectively), m can be expected

only a relatively small additional reduction of correlation between neighboring pilot signals occurs when T is increased to a value greater than 5ps. m [0050] Tables 3 and 4 show the result of the Doppler spread estimation for the OFDM channel, in the case of estimates made on 177 continuous pilot signals, the results of which are then combined, for the case SNR = 10 dB and 30 dB, respectively.

<td>Current f = D</td><td>5 Hz</td><td>10 Hz</td><td>20 Hz</td><td>50 Hz</td><td>100Hz</td>

<td>dispersion delays</td><td></td><td></td><td></td><td></td><td></td>

<td>T = 1ps</td><td> 17(3.0)</td><td> 22 (3.1)</td><td> 31 (3.2)</td><td> 60 (9.8)</td><td> 106(17)</td>

<td>T = 5ps</td><td> 18(2.2)</td><td> 22 (1.4)</td><td> 32 (2.2)</td><td> 59 (5.8)</td><td> 105(10)</td>

<td>T = 10ps</td><td> 18(1.8)</td><td> 23(1.9)</td><td> 32 (2.3)</td><td> 62 (4.5)</td><td> 105 (9.0)</td>

<td>T = 20ps</td><td> 18(1.8)</td><td> 22 (1.3)</td><td> 32 (1.9)</td><td> 60 (3.6)</td><td> 104 (8.9)</td>

Table 3: Comparison of the mean and standard deviation for estimating Doppler spread under different current Doppler spread conditions and different maximum delay spreads for a measurement time of 100 ms and SNR = 10 dB

<td>Current f = D</td><td>5 Hz</td><td>10 Hz</td><td>20 Hz</td><td>50 Hz</td><td>100Hz</td>

<td>delay dispersion</td><td></td><td></td><td></td><td></td><td></td>

<td>T = 1ps</td><td> 7.1 (1.0)</td><td> 12 (1.9)</td><td> 23 (4.2)</td><td> 51 (11.2)</td><td> 96(19)</td>

<td>T 5ps m =</td><td> 6.8 (0.7)</td><td> 12 (1.4)</td><td> 22 (2.1)</td><td> 52 (5.5)</td><td> 98 (11)</td>

<td>T 10ps m =</td><td> 6.8 (0.6)</td><td> 12 (1.9)</td><td> 22 (2.0)</td><td> 52 (5.5)</td><td> 97 (8.6)</td>

<td>T 20ps m =</td><td> 7.0 (0.5)</td><td> 12(1.3)</td><td> 22 (1.7)</td><td> 51 (5.1)</td><td> 96 (8.3)</td>

Table 4: Comparison of the mean and standard deviation for estimating Doppler spread under different current Doppler spread conditions and different maximum delay spreads for a measurement time of 100 ms and SNR = 30 dB [0051] Comparing the results shown in Tables 1 and 2 (where Doppler scatter on a given OFDM channel was estimated by considering only one carrier frequency), there is a clear improvement in results. The following have been particularly noted:

• A significant improvement was obtained in estimating the Doppler extension in a given OFDM channel using a combination of measurements from a number of carrier frequencies compared to an estimation of the Doppler extension made using only one frequency.

Considering the case f = 5 Hz and 10 dB SNR, the reduction of the standard deviation is

D

25 / 1.8; 14 which shows good agreement with the theory.

• The results do not improve when the Doppler delay extension in a given channel, T increases m

above 5ps for the specific example under consideration.

• Assuming that the spread of delays is large enough to consider the exceeding of the zero value for different frequencies used in the Doppler estimation to be uncorrelated, the estimation time may be reduced in proportion to the number of carrier frequencies used in the estimation. This is because the accuracy of the Doppler estimate is essentially determined by the number of exceedances of the zero value used to make the estimate. To achieve greater accuracy, a large number of zero exceedances is necessary. If the Doppler estimate is made using only one carrier frequency, it means that the estimation time must be increased to obtain a sufficient number of zero exceedances. However, if, say, N carrier frequencies were used when estimating the Doppler effect, then N times the number of exceedances of zero value can be expected, compared to the case in which only one frequency was used. Consequently, the same number of exceedances of the zero value can be expected in the estimation method using N carrier frequencies if the estimation time is reduced in the same proportion N.

• In most cases, the improvement in accuracy using all continuous pilot signals is associated with a reduction in measurement time in excess of 100 for the specific examples discussed.

[0052] FIG. 8 is a flowchart of the generalized procedure used to determine Doppler spread in a given OFDM channel. The procedure may be performed, for example, in a team estimating Doppler spread in an OFDM receiver. For example, if the exemplary OFDM receiver of FIG. 2 is intended to be modified so that it ultimately has logic circuits to implement the invention, the logic circuit used to determine the Doppler spread in a given OFDM channel in accordance with the invention can be implemented as a component of channel estimator 213.

[0053] To estimate the Doppler extension on a given OFDM channel, a set of carrier frequencies at which the Doppler extension will be measured will be determined (step 80). For example, in OFDM systems that transmit pilot signals such as DVB-T and DVB-H, some or all carrier frequencies containing continuous pilot signals may be selected. In some exemplary embodiments, the number of carrier frequencies to be measured is constant. In alternative embodiments, the number of carrier frequencies to be measured changes dynamically, it can be based, for example, on the amount of delay spread in a given channel. The techniques used to estimate the spread of delays on a given OFDM channel are well known, and there is no need to describe them in detail here. For example, the spread of delays can be determined in accordance with US Patent Application No. 11/110840, entitled "Initial parameter estimation in OFDM systems" Anders Berkeman et al. Filed April 21, 2005.

[0054] Using a specific set of measurement frequencies, a Doppler spread estimate is made for each of these carrier frequencies, resulting in a set of Doppler spread estimates (step 803). These estimates can be made using any known techniques used to estimate Doppler scattering on a single carrier frequency, including any of the techniques previously described. For example, any of the techniques mentioned above that involve zero crossing or level exceeding can be used, with or without hysteresis, as previously described.

[0055] As explained before, many factors can affect the accuracy of measurements, including the length of measurement time T and the range over which the carrier frequencies at which meas' measurements are performed are correlated with each other. Also, as explained earlier, the scope of correlation depends on the delay spread in a given channel. These relationships can be used in alternative, exemplary embodiments by adjusting the measurement time T as a function of the magnitude of meas, the delay dispersion specified for the OFDM channel under consideration. More specifically, exemplary embodiments may be derived from dynamically regulating measurement time solutions as a function of the amount of delay spread in a given channel. Since the higher the level of delay spread, the smaller the correlation between the measurement frequencies belonging to the set will be, the desired level of accuracy can be achieved with a shorter measurement time when the delay spread is greater.

[0056] Accordingly, in some example embodiments, the measurement time may have lower values ​​as the delay spread over the channel is high. In yet other alternative embodiments, the measurement time T may be selected from a limited meas set of measurement times (e.g., from two possibilities) based on whether the delay spread value is at or above a predetermined level, determined empirically based on the supplied system characteristics . For example, in the example system previously described, it was shown that the frequencies belonging to the set of transmitting continuous pilot signals are substantially uncorrelated with each other when the delay spread is at the level of T = 5ps or above. Therefore, it is possible to design a system in which relatively long time measurement intervals T are used when the delay spread occurring in a given channel meets the condition T &lt;5ps, and relatively short time measurement intervals m

T when the delay spread occurring in a given channel meets the condition T&gt; 5ps, meas m

Of course, the purpose of these values ​​and threshold conditions is to provide an example; the current threshold values ​​and measurement conditions will have to be adjusted to the conditions prevailing in the channels of the particular system, they are expected to occur in any specific implementation.

[0057] The unit Doppler spread estimation specified in step 803 is then combined to produce a Doppler spread estimation occurring on the given OFDM channel (step 805). The connection can be carried out, for example, by averaging the Doppler scatter estimates determined for individual channels followed by using this average value as an estimation of the Doppler scatter present on a given OFDM channel. Simple averaging techniques can be used, or in alternative embodiments, weighted averages can be used in which the weight of any of the Doppler spread estimates is a function of signal quality on individual carrier frequencies.

[0058] The invention has been described with reference to specific embodiments of an exemplary nature. However, from those skilled in the art, it is clear that it is possible to carry out the invention using specific methods other than those described above.

[0059] For example, a method of creating a set consisting of two or more carrier frequencies selected from a plurality of carrier frequencies has been described, wherein their number N is a variable value where N is a function of the value representing the delay extension occurring on a given channel. A method of estimating the Doppler spread value for each of the selected carrier frequencies may also be described, which may include a measurement carried out on each of the carrier frequencies in the T time interval, the length of the T time interval being a function of the meas meas value representing the value of the delay spread in a given channel. All these aspects must be combined and implemented in each of the specific performances. For example, based on the delay spread value in a given channel, an appropriate N value can be determined. Then, based on a fixed N value, an appropriate time period value can be determined

T, so as to obtain sufficient accuracy of measurements. Since in this case T is a meas meas function of N, it thus also becomes a function of the delay spread value.

[0060] Thus, the described embodiment is only an example and should not be regarded as limiting in any way. The scope of the invention is defined to a greater extent by the following claims than by the above description, it is the intention of any modifications and equivalent solutions that fall within the scope of the claims to be encompassed by the following claims.