WO2014106540A1

WO2014106540A1 - Method for estimating and removing dc offset

Info

Publication number: WO2014106540A1
Application number: PCT/EP2013/050069
Authority: WO
Inventors: Sha HU; Gengshi Wu; Basuki Endah Priyanto; Shashi Kant; Jianjun Chen
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2013-01-03
Filing date: 2013-01-03
Publication date: 2014-07-10
Also published as: EP2941852A1; CN104904172A; CN104904172B

Abstract

The present invention relates to a method in a direct conversion receiver for estimating and removing DC offset, said direct conversion receiver being arranged to receive radio communication signals in a wireless communication system, the method comprising the steps of: receiving and down sampling time domain signals, wherein said time domain signals comprise at least one first synchronisation signal; estimating a residual DC offset Δd based on said at least one first synchronisation signal; computing a refined DC offset estimation d̃ based on said estimated residual DC offset Δd; and applying said refined DC offset estimation d̃ on said time domain signals so as to remove DC offset from said time domain signals. Furthermore, the invention also relates to a receiver device, a computer program, and a computer program product thereof.

Description

METHOD FOR ESTIMATING AND REMOVING DC OFFSET

Technical Field

The present invention relates to a method for estimating and removing DC offset in a direct conversion receiver. Furthermore, the invention also relates to a receiver device, a computer program, and a computer program product thereof.

Background of the Invention

Direct Current (DC) offset is a major problem in direct conversion receivers. A part of the Local Oscillator (LO) signal is down converted to a baseband signal which leads to unwanted DC offsets (see e.g. Rainer Stuhlberger, R. Krueger, B. Adler, J. Kissing, L. Maurer, G. Hueber and A. Springer, " LTE-Downlink Performance in the Presence of RF-Impairments", Proceedings of the 10th European Conference on Wireless Technology, pp. 189-192, October 2007).

If the DC offset is not properly handled, it can affect both cell search and data demodulation performance. For example, the cell search in Long Term Evolution (LTE) systems requires time domain correlation and unwanted DC offset can therefore destroy the correlation property of cell search synchronization signal. Although the DC sub-carrier is not used for data transmission in LTE systems, the Channel State Information (CSI) estimation around the DC component will be affected by the DC offset that directly impacts the performance of LTE system. It is thus very important to estimate and compensate the DC offset before the data demodulation and decoding at the receiver side.

In LTE systems, the DC offset compensation methods mainly involves two prior art approaches, i.e.:

• Notch filter or high pass filter approach - the notch filter or high pass filter is a digital filter circuitry that provides a notch at the frequency corresponding to the DC offset frequency and thus can remove the DC offset component from the received signal;

• Time domain received samples averaging approach - this method involves averaging received samples to estimate the DC offset. The notch filter or the high-pass filter approach requires a long transient time and it will distort the noise power spectrum density which adds complexity to whiten the distorted noise. Also in LTE, the DC offset unit may need to support different bandwidth configurations which make the digital filter design more complex.

As for the averaging method under low Signal to Noise Ration (SNR) region and fading scenario, the DC offset estimation through sample averaging is inaccurate and may degrade the overall performance substantially. From the above it is concluded that in LTE system and other relevant communication systems the DC offset distorts the time domain received samples and therefore degrades the system performance. Hence, it is very important to get an accurate DC offset estimation and compensate for the DC offset at the receiver using the DC offset estimation to not degrade system performance.

Summary of the Invention

An object of the present invention is to provide a solution which mitigates or solves the drawbacks and problems of prior art solutions. Another object of the invention is to provide an improved solution to the problem of estimating DC offset and mitigating the same.

According to a first aspect of the invention, the above mentioned objects are achieved by a method in a direct conversion receiver for estimating and removing DC offset, said direct conversion receiver being arranged to receive radio communication signals in a wireless communication system, the method comprising the steps of:

- receiving and down sampling time domain signals, wherein said time domain signals comprise at least one first synchronisation signal;

- estimating a residual DC offset Ad based on said at least one first synchronisation signal; - computing a refined DC offset estimation d based on said estimated residual DC offset Ad ; and - applying said refined DC offset estimation d on said time domain signals so as to remove DC offset from said time domain signals.

Preferred embodiments of the method above are defined in the appended dependent claims.

The present method can also be executed in processing means of e.g. a computer. The method may be comprised in a computer program product.

According to a second aspect of the invention, the above mentioned objects are also achieved with a direct conversion receiver device arranged to receive radio communication signals in a wireless communication system; the direct conversion receiver device further being arranged to:

- receive and down sampling time domain signals, wherein said time domain signals comprise at least one first synchronisation signal;

- estimate a residual DC offset Ad based on said at least one first synchronisation signal;

- compute a refined DC offset estimation d based on said estimated residual DC offset Ad ; and

- apply said refined DC offset estimation d on said time domain signals so as to remove DC offset from said time domain signals.

The direct conversion receiver device may be modified so as to correspond to all different embodiments of the present method.

The present invention provides an improved DC offset estimation method for direct conversion receivers which result in improved performance at the receiver. This is achieved by using the first synchronisation signal for estimation of a residual DC offset which in turn is used for computing a refined DC offset estimation having higher accuracy than prior art solutions. Further, according to preferred embodiments the present method jointly estimates the time domain CSI of the first synchronisation signal and the residual DC offset at the same time. Thereby, the accuracy of CSI estimation of the first synchronisation signal can be improved. The cell search procedure accuracy and Down Link (DL) data performance is also improved by the invention. Thus, complicated notch filter or high-pass filter approaches which need to support different bandwidth configurations or different modes in a multimode modem is also avoided with the present method.

Further applications and advantages of the invention will be apparent from the following detailed description.

Brief Description of the Drawings

The appended drawings are intended to clarify and explain different embodiments of the present invention in which:

Fig. 1 illustrates a direct conversion receiver structure;

Fig. 2 shows the DL SCH structure in LTE FDD mode;

Fig. 3 illustrates the cell search procedure in LTE systems at the receiver side;

Fig. 4 illustrates an embodiment of a receiver structure according to the present invention;

Fig. 5 illustrates a residual DC offset estimation structure; and

Fig. 6 illustrates the flow of DC offset estimation and compensation method according to an embodiment of the present invention.

Detailed Description of the Invention

To achieve the aforementioned and further objects, the present invention relates to a method in a direct conversion receiver for estimating and removing DC offset. As mentioned above, the system performance can be substantially improved by providing accurate DC offset estimations.

According to preferred embodiments, the estimation and removal of DC offset can also be combined with cell search procedures, such as determining frame head, Cell ID, Cyclic Prefix (CP) type, etc. Thereby, reduced computational complexity when determining DC offset estimation can be provided. The cell search procedure accuracy and DL data performance is also improved. In the following the DL synchronisation procedures and parameters of LTE systems are described so as to provide a basis for the further detailed explanation of the present invention.

In LTE systems, one radio frame duration is 10 ms long and consists of 10 subframes (numbered from 0 to 9), and each subframe can be divided into two slots. Thus, each radio frame includes 20 slots numbered from 0 tol 9. Further, each slot includes several Orthogonal Frequency-Division Multiplexing (OFDM) symbols depending on the CP type and subcarrier configuration (e.g. one slot consists of 7 OFDM symbols per slot in normal CP type with subcarrier spacing of 15 kHz).

The DL Synchronization Channel (SCH) in LTE is transmitted every 5 subframes in the centre/middle 6 Resource Blocks (RBs) regardless of the bandwidth configuration. The SCH includes two parts, namely: the Primary Synchronization Signal (PSS) and the Secondary Synchronization Signal (SSS). The SCH is transmitted at different position in the radio resources depending on the frame structure, i.e. Frequency Division Duplex (FDD) or Time Division Duplex (TDD) mode.

In FDD mode the PSS is mapped on the last OFDM symbol in slot 0 and 10, and the SSS is mapped on the previous OFDM symbol adjacent to the OFDM symbol in which the PSS is mapped. A typical FDD DL SCH RB mapping is shown in figure 2. On the other hand, in TDD mode the PSS is mapped to the third OFDM symbol in subframes 1 and 6 and the SSS is mapped to the last OFDM symbol in slot 1 and 1 1.

Furthermore, in LTE the PSS sequence p{n) is generated from a frequency-domain Zadoff- Chu sequence according to:

The root u is decided by the cell sector ID N_jQ which can have three possible values: 0, 1 or 2. The mapping relationship between the root u and the cell sector ID NJ_Q is defined in as follows: iV⁽²⁾ u

V ID

0 25

1 29

2 34

An FTT with N=128 point can transform the frequency domain 62-length PSS sequence to time domain 128-length PSS sequence:

where p_u (k) is the fre uency domain PSS sequence with zero padding:

Λ(*) Eq. (3).

Moreover, in LTE systems the cell search procedure relies on the SCH. Firstly, PSS detection is applied on down sampled time domain samples (with sampling rate of 1.92 Mbps) to get the slot start position through a slicing correlation with possible local PSS sequences. After the time domain PSS position is acquired, a successive SSS detection is applied in the frequency domain to detect the frame head, Cell ID and CP type. The SSS detection is based on the correlation between received SSS frequency domain samples and all possible local SSS sequences at the receiver. In order to remove the channel affect from the received SSS samples before the SSS detection, the CSI estimation of the PSS can be used to conjugate multiply with the extracted SSS sequence before implementing a FFT to transform the SSS sequence into the frequency domain. Figure 3 illustrates the above described cell search structure.

The present method in a direct conversion receiver involves receiving and down sampling time domain signals which are received from a transmitter in the wireless communication system. The time domain signals should comprise at least one first synchronisation signal for synchronisation of received data. Thereafter a residual DC offset Ad based on the at least one first synchronisation signal is estimated and a refined DC offset estimation d is computed based on the estimated residual DC offset Ad . Finally, the refined DC offset estimation d is applied on the time domain signals so as to remove DC offset from the time domain signals. Thereby, improved performance is provided by the present method. According to an embodiment of the invention the method also comprises the steps of: estimating a raw DC offset d based on the down sampled time domain signals; subtracting the estimated raw DC offset d from the down sampled time domain signals; and detecting the first synchronisation signal based on the subtracted down sampled time domain signals. This means that the first synchronisation signal is detected based on the down sampled time domain signals which have been subtracted with the raw DC offset d . Preferably, the first synchronisation signal is the PSS in LTE systems, which also means that a second synchronisation signal is the SSS for mentioned types of systems. The PSS and SSS are transmitted in the DL in the SCH. Furthermore, according to yet another embodiment of the invention the present method further comprises the steps of: jointly estimating the residual DC offset Ad and CSI in the subtracted down sampled time domain signals based on the detected first synchronisation signal. The advantages with this approach have been clearly stated earlier. To understand the present invention more thoroughly a mathematical model of received radio signals is also presented in the following disclosure. The system parameters are in this example from LTE systems but the skilled person realises that these values can vary depending on the specific wireless communication system in which the present method is implemented.

For the linear model, at the receiver side, time domain samples can be modelled as,

L-l

y[n] =∑h[l, n]x[n - l] + d + vin] _Eq. ₍₄₎

/=o

where xfnj is the transmitted time domain signal, h[l, n] (1=0, 1 , ... , L-l ) is the channel taps at sampling time n and L ( L≥ I) is the channel time delay, d is the DC offset to be estimated and w[n] is the thermal noise. After accumulation over the down sampled (1.92 Mbps) time domain samples in a specific period a raw estimation d of the actual DC offset / is obtained:

where M is the total samples used for estimation of the raw DC offset d for the specific averaging time. In every 5 ms, maximal M = 1.92 MHz x 5ms = 9 600 samples can be used for the raw estimation in a LTE system. The raw estimation is thus an initial DC offset estimation. While when the received signal suffers from frequency selective fading which is common in wireless communication environment or when the SNR is low, the raw estimation may have a fairly large estimation error that results in a residual DC offset remaining in the received signals after removing the raw DC offset estimation from the received signals.

The PSS (the first synchronisation signal) based residual DC offset estimation according to an embodiment is given as follows: after subtracting the raw estimation d from the received samples in Eq. (4), the time domain PSS signal can be modelled as,

n = k +L - \,k + L,k + L + l,...,k + L + m - 2

where PSS[ri] is the time domain PSS signal, h[l, n] is the l-t channel tap at sampling time n and L is the channel time delay, d is the DC offset, wfnjis the thermal noise, and m is the PSS length used for the residual DC offset estimation. Since h[l,n] (1=0, 1, ... , L-l) can be assumed constant over one OFDM symbol in the LTE system, n is omitted from h[l,n] in the rest part and Eq. (5) is rewritten as

Further, k = k + k_offset is the start position index of received samples used for the DC offset estimation, k is the detected PSS start position and k_offset is a chosen offset in order to only utilize part of the received PSS signals. For instance, if the whole received PSS samples y[n] are used (but discard the samples affected by the multi-path channel in the start positions, which means indices n= k+L-1, k+L, ... , k+L+m-2) for the DC offset estimation, in this case k - k , k_offset - 0, m = 128 - (L - 1) , and m = 128 - (L - 1) ; while if only the middle 64 received

PSS signals y[n] are used (indices n= k+32, k+33, ... , k+95) in this case, k =k + 33-L,

In general, the use of longer sequences for estimation of the residual DC offset will render a better performance but require a larger memory for coefficients storage. Denote

Ad— d— d

Y[m] = (y[k+L-l],y[k+L],...,y[k+L + m 2]f

PSS[k_offset +L-1] PSS[k_offset +L-2]

PSS[k_offsei+L] PSS[k_offsel+L-l]

P[m] = PSS[k_offset+L ₊ \] PSS[k_offsel+L]

PSS[k_offsel+L + m-2] PSS[k_osel+L + m-3]

H[L] = (h[0],h[\],...,h[L-l])

H[L, Ad] = (h[0], h[\],..., h[L - l],Ad)^r - (H[L], Adf

W[m] = [^k +L-\],w k +L],...,M k +L + m-2]

The Eq. (5) can then be written in matrix form as,

Y[m] = P[m]H[L, d] + W[m] Eq. (6)

where H[L,Ad] is a column vector containing the CSI vector H[L] and the residual DC offset Ad. Assuming that W[m] is Additive White Gaussian Noise (AWGN), the ML estimation of H[L,Ad] leads to the LS solution, i.e.,

H[L, Ad] - (P^H [m]P[m]Y P^H[m]Y[m] - C[m]Y[m] Eq. (7)

where,

C[m] = (P"[m]P[m])¹ P^H[m] Eq. (8).

Since P[m] is a known mx(L+\) matrix based on the local time domain PSS sequences that are transformed from local frequency domain PSS sequences with 128-length FFT as described in Eq. (1) and Eq. (2), C[m] is thus a (L + Y)xm matrix which can be obtained from matrix P[m] as described in Eq. (8). For all (possible) three PSS sequences

(2)

corresponding to different cell sector IDs N_ID , different matrix C[m] can be pre-calculated and stored. For residual DC offset estimation, the corresponding matrix C[m] is chosen

(2)

depending on the cell sector ID N_ID detection result in initial cell search (or in periodic cell search which is already known). Then the joint PSS CSI and residual DC offset estimation method for H[L, Ad] described in Eq. (7) is just a simple matrix-multiplication between the chosen matrix C[m] and the received time domain PSS samples Y[m] from which the raw estimation d has been removed according to Eq. (5). As residual DC offset Ad is taken into consideration for the CSI estimation H[L] the joint estimation H[L, Ad] renders a better estimation for the PSS CSI.

Example

Assuming channel taps L=9 (EVA or ETU channels defined in 3 GPP LTE) and using the middle 64 received PSS samples for DC offset estimation; in this casQ k_offset = 33 - L = 24 and m = 64.. For different cell sector IDs based on Eq. (2) and Eq. (3), local time domain PSS sequence PSS(n), n = 0,1,...,127 can be generated, and [m] is a 64x l0 matrix including local PSS signals from index 24 to 95,

"PSS132] PSS[3 \] . . PSS[24]

PSS[33] PSS[32] . . PSS[25] 1

P[m] = PSS[34] PSS[33] . . PSS[26] 1 PSS[95] PSS[94] . . PSS[87] 64x10

The coefficients C[m] are in the form of 10x 64 matrix which can be calculated and stored for all cell sector IDs .

Regarding the refined DC offset d this offset can according to an embodiment be computed by adding the estimated raw DC offset d and the estimated residual DC offset Ad which means that the refined offset is given by d = d + Ad . This is essential, especially when the SNR is low or the channel is frequency selective fading, the estimation error of raw DC offset d obtained from averaging the received time domain samples results in a fairly large residual DC offset and thus degrade the performance. Therefore the refined DC offset d is a better estimation than the raw DC offset d .

As mentioned above, in LTE systems the SCH is always transmitted in the middle/central 6 RBs every 5 ms in the LTE/LTE-A downlink and the UE will implement the cell search procedure periodically. Therefore, the PSS can be used to estimate the DC offset. Hence, according to an embodiment of the invention the present method can be implemented in the following steps:

1. Average Low Pass (LP) filtered samples at 1.92 Mbps sampling rate that is down sampled from the original received time domain signals to get a raw estimation of the DC offset d ;

2. Subtract the raw estimation DC offset d in step 1 from the LP filtered samples and apply PSS detection;

3. After acquired the PSS position, apply a joint CSI and residual DC offset estimation Ad based on the extracted 128-length time domain PSS signals;

4. Subtract the residual DC offset estimation Ad from the LP filtered samples before the SSS (the second synchronisation signal) detection and other successive cell search procedures;

5. Add the raw estimation of the DC offset d in step 1 and the residual DC offset estimation Ad in step 3 together to obtain a refined DC offset estimation for data demodulation of the received signals;

6. Apply an alpha filter to further de-noising the refined estimation obtained in step 5;

7. Compensate the DC offset from the original time domain samples before data demodulation and decoding.

This embodiment is also illustrated in a receiver device structure in figure 4 and as a flow chart in figure 6. It is noted from figure 4 that the DC offset estimation device in figure 4 is combined with a cell search unit. Every 5 ms the SCH will be transmitted in the middle 6 RBs and the cell search procedure will start based on the down sampled received samples y[k] . Firstly, averaged the down sampled signals for a specific period in every 5 ms (maximal 5 ms) at 1.92 M sampling rate to get a raw estimation of DC offset, i.e. d . Then subtract the raw DC offset d from the down sampled signals and start PSS detection: i.e. y[k] = y[k] - d .

If in initial cell search slicing correlation is needed with all three possible local PSS sequences while in a periodic search only with one known PSS sequence. After detecting the PSS position and the corresponding sequence, the residual DC offset Ad and PSS CSI can be jointly detected. Then the residual DC-Offset Ad will be further removed from the samples before SSS detection and other cell search procedures: y[k] = y[k] - Ad . The residual DC offset will also be combined with the raw estimation to obtain the refined estimation d : d = d + Ad . The refined DC offset estimation will be passed through an alpha filter before being used to compensate for the DC offset before demodulation and decoding. The alpha filter will further improve the DC offset estimation and therefore also the performance.

The residual DC offset and CSI joint detection unit is described in the following text with reference to figure 5. The three possible filter matrices CfmJ are pre-computed and stored in a memory according to a pre-defined filter length m. After the PSS position and the cell sector

(2) (2)

ID N_ID have been detected (in periodic cell search N_ID is known): firstly extract the 128- length PSS sequence from the down sampled signals; then, choose the right filter matrix CfmJ and obtain the joint estimation H[L,Ad] using Eq. (7): H[L, Ad] = (x^H[m]X[m] ¹ X^H[m]Y[m] = C[m]Y[m] .

For cell search procedure, further subtract the residual DC offset estimation Ad from y[k] and implement the SSS detection and further cell search procedures. The estimated PSS CSI can be used as the initial CSI of SSS to remove the channel affect before SSS detection.

Furthermore, as understood by the person skilled in the art, any method according to the present invention may also be implemented in a computer program, having code means, which when run by processing means causes the processing means to execute the steps of the method. The computer program is included in a computer readable medium of a computer program product. The computer readable medium may comprises of essentially any memory, such as a ROM (Read-Only Memory), a PROM (Programmable Read-Only Memory), an EPROM (Erasable PROM), a Flash memory, an EEPROM (Electrically Erasable PROM), or a hard disk drive. Moreover, the present invention also relates to a direct conversion receiver device corresponding to all embodiments of the present method. The receiver device comprises all necessary means and is arranged to perform the method according to the present invention which means that the device may e.g. include: processing means, signal input means, signal output means, sampling means, memory means, communication means, etc. The present invention is according to an embodiment used in a 3 GPP wireless communication system such as LTE or LTE Advanced. Hence, in this case the receiver device is a part of, or comprised in a user equipment (UE).

Finally, it should be understood that the present invention is not limited to the embodiments described above, but also relates to and incorporates all embodiments within the scope of the appended independent claims.

Claims

1. Method in a direct conversion receiver for estimating and removing DC offset, said direct conversion receiver being arranged to receive radio communication signals in a wireless communication system, the method comprising the steps of:

- estimating a residual DC offset Ad based on said at least one first synchronisation signal;

- computing a refined DC offset estimation d based on said estimated residual DC offset Ad ; and

- applying said refined DC offset estimation d on said time domain signals so as to remove DC offset from said time domain signals.

2. Method according to clam 1, further comprising the step of:

- estimating a raw DC offset d based on said down sampled time domain signals;

- subtracting said estimated raw DC offset d from said down sampled time domain signals; and

- detecting said at least one first synchronisation signal based on said subtracted down sampled time domain signals.

3. Method according to claim 2, further comprising the steps of:

- jointly estimating said residual DC offset Ad and channel state information (CSI) in said subtracted down sampled time domain signals based on said detected at least one first synchronisation signal.

4. Method according to claim 3, further comprising the step of:

- subtracting said estimated residual DC offset Ad from said subtracted down sampled time domain signals.

5. Method according to claim 2, wherein said refined DC offset d is computed by adding said estimated raw DC offset d and said estimated residual DC offset Ad , i.e. d = d + Ad .

6. Method according to claim 5, further comprising the step of:

- applying an alpha filter on said refined DC offset so as to de-noising said refined DC offset d .

7. Method according to claim 3, wherein said time domain signals further comprises at least one second synchronisation signal; and said method further comprises the steps of:

- detecting said at least one second synchronisation signal; and

- performing further cell search procedure steps based in said down sampled time domain signals.

8. Method according to claim 7, wherein said wireless communication system is a 3 GPP cellular system such as LTE or LTE Advanced; and said at least one first synchronisation signal is a primary synchronisation signal (PSS) and said at least one second synchronisation signal is a secondary synchronisation signal (SSS).

9. Computer program, characterised in code means, which when run by processing means causes said processing means to execute said method according to any of claims 1-8.

10. Computer program product comprising a computer readable medium and a computer program according to claim 9, wherein said computer program is included in the computer readable medium, and comprises of one or more from the group: ROM (Read-Only Memory), PROM (Programmable ROM), EPROM (Erasable PROM), Flash memory, EEPROM (Electrically EPROM) and hard disk drive.

11. Direct conversion receiver device arranged to receive radio communication signals in a wireless communication system; the direct conversion receiver device further being arranged to: - receive and down sampling time domain signals, wherein said time domain signals comprise at least one first synchronisation signal;

12. Direct conversion receiver according to claim 11, wherein said wireless communication system is a 3 GPP cellular system such as LTE or LTE Advanced, and said direct conversion receiver device is comprised in a user equipment (UE).