US20090202021A1

US20090202021A1 - Frequency offset compensation for detecting random access channel prefix

Info

Publication number: US20090202021A1
Application number: US12/239,989
Authority: US
Inventors: Fang-Chen Cheng; Lei Song
Original assignee: Individual
Current assignee: Alcatel Lucent SAS; Nokia of America Corp
Priority date: 2007-10-01
Filing date: 2008-09-29
Publication date: 2009-08-13
Also published as: WO2009045734A3; TWI469608B; WO2009045734A2; US20090196238A1; US9319195B2; US20090196370A1; US8351532B2; TW200939722A

Abstract

An exemplary method of communicating includes shifting a constant-amplitude zero autocorrelation (CAZAC) root sequence to a shifted CAZAC sequence. The CAZAC root sequence is used by a source of a received communication. The shifted CAZAC sequence is used for detecting a preamble of the received communication.

Description

CROSS REFERENCE TO RELATED APPLICATIONS

This application claims priority to U.S. Provisional Application No. 60/997,115 which was filed on Oct. 1, 2007.

1. FIELD OF THE INVENTION

This invention generally relates to communication. More particularly, this invention relates to compensating for frequency offsets in communications.

2. DESCRIPTION OF THE RELATED ART

Wireless communication systems are well known and in widespread use. Typical cellular communication arrangements include a plurality of base station transceivers strategically positioned to provide wireless communication coverage over selected geographic areas. A mobile station (e.g., notebook computer or cellular phone) communicates with a base station transceiver using an over-the-air interface. The communications from a mobile station to the base station may be affected by movement of the mobile station. For example, when a mobile station is moving at high speed, a Doppler effect introduces a frequency offset into the signaling from the mobile station in an uplink direction.
Random access channel (RACH) communications have a preamble format with a cyclic prefix to enable frequency domain processing. The current preamble structure includes a 0.1 ms long cyclic prefix and a 0.8 ms long main part. A cyclic prefix of 0.1 ms is sufficient for cells up to 15 km. For larger cells, however, such a cyclic prefix is too short and simple frequency domain processing is not possible.
For the generic frame structure, random access occupies a bandwidth of 1.08 MHz (6 resource blocks) and its length is a multiple of a 1 ms transmit time interval. The location in the frequency domain is controlled by the parameter k₀, configured by higher layers in multiples of N_BW ^RBand fulfilling 0≦k₀<N_BW ^UL−N_BW ^RB.
The random access preambles are generated from Zadoff-Chu sequences with zero correlation zone (ZC-ZCZ) generated from one or several root Zadoff-Chu sequences. The network configures the set of preamble sequences that each mobile station is allowed to use.
The u^throot Zadoff-Chu sequence is defined by
$x_{u} (n) = e^{- j \frac{π un (n + 1)}{N_{ZC}}}, 0 \leq n \leq N_{ZC} - 1$
where the length N_ZCof the Zadoff-Chu sequence is equal to 839. From the u^throot Zadoff-Chu sequence, random access preambles with zero correlation zone are defined by cyclic shifts of multiples of N_CSaccording to
x _u,v(n)=x _u((n+vN _CS)mod N _ZC)
where N_CSis configurable by the upper layer. According to one standard, N_CSmay have 16 possible values.
RACH preamble detection is challenging when a mobile station is moving fast because the high speed may result in a high frequency offset. The maximum frequency offset f_offset,ULseen at the base station receiver (e.g., eNB) is obtained as
f _offset,UL =Δf _BS +Δf _UE+2×f _Doppler _— _max
where Δf_BS, Δf_UE, and f_Doppler,maxdenote the base station frequency drift, mobile station frequency error, and the maximum Doppler frequency, respectively.
In some example systems, the frequency accuracy requirements at the base station are 0.05 ppm and 0.1 ppm of the carrier frequency. For a carrier frequency of 2.1 GHz, the maximum tolerable frequency offset is 781 Hz for a mobile station moving at 120 Km/h. The worst-case frequency offset is 1675 Hz when the speed is 350 Km/h.
Rach preamble detection is done in effect by a constant amplitude zero autocorrelation (CAZAC) sequence correlator implemented in the frequency domain. Two issues need to be taken care of when operating in high frequency offset environment. One is that the frequency offset may result in additional correlation peaks in the preamble detection in the time domain.
Separating the dominant additional peak from the normal correlation peak depends on the root index of the CAZAC sequence. The effect of the additional correlation peaks can be reduced by sequence restrictions meaning that the set of available root sequences and their cyclic shifts are limited in such a way that false detections due to the additional peaks overlap in time with other shifts' peaks can be avoided by only allocating a subset of otherwise N_zcavailable shifts.
Frequency offsets also affect the detection performance and false alarm rate when the baseline preamble sequence is CAZAC with circular shifts. This effect is recognizable when considering the inter-subcarrier interference (ISI) with CAZAC sequence. Assume that the CAZAC sequence is directly loaded on the usable subcarriers for RACH. Then, each subcarrier conveys one chip of the CAZAC sequence. Any frequency offset at the receiver (eNB) due to Doppler spread or residual frequency offset results in the frequency sampling position not being aligned with subcarrier position. The result is a mixed signal with neighbor subcarriers. In some cases, may be on the order of 3.5 dB.
One suggested approach to resolving these issues is to use a shorter preamble length for high speed situations. With this approach, the detection performance and false alarm is acceptable with 2 times repetition in 1 ms RACH. One drawback to this approach is that it limits the number of the available root ZC sequences and Zero-Correlation Zone (ZCZ) sequences because of the short sequence length. The result is that it becomes difficult to plan large cells.
Another suggested approach uses the current 0.8 ms preamble and implements shorter coherent correlation by partitioning the whole sequence into several segments. Each segment will perform correlation with segmented reference signal and later be combined non-coherently. While this method results in better performance in high frequency offset situations, it nonetheless incurs significant implementation complexity and performance degradation for low frequency offset users. Therefore, this approach is not likely to be accepted as a solution.

SUMMARY

An exemplary method of communicating includes shifting a constant-amplitude zero autocorrelation (CAZAC) root sequence to a shifted CAZAC sequence. The CAZAC root sequence is used by a source of a received communication. The shifted CAZAC sequence is used for detecting a preamble of the received communication.
An exemplary receiver device comprises a detector module that is configured to shift a CAZAC root sequence to a shifted CAZAC sequence. The CAZAC root sequence is used by a source of a communication received at the receiver device. The shifted CAZAC sequence is used for detecting a preamble of the received communication.
The various features and advantages of the disclosed examples will become apparent to those skilled in the art from the following detailed description. The drawings that accompany the detailed description can be briefly described as follows.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 schematically illustrates selected portions of a communication network designed according to an embodiment of this invention.

FIG. 2 schematically illustrates a feature of an example embodiment.

FIG. 3 is a flow chart diagram summarizing one example approach.

DETAILED DESCRIPTION

FIG. 1 shows selected portions of a wireless communication system 20. A base station 22 communicates with a mobile station 24. The communications between the base station 22 and the mobile station 24 occur in a downlink direction and an uplink direction. In this example, the mobile station 24 uses at least a random access channel (RACH) for communications in the uplink direction.
At times the mobile station 24 will be moving at a relatively high speed. Under such circumstances, the receiver of the base station 22 compensates for frequency offset introduced by a Doppler effect associated with the high speed of the mobile station 24. A baseband processor of the base station 22 shifts a constant-amplitude zero autocorrelation (CAZAC) root sequence to a shifted CAZAC sequence. The CAZAC root sequence is used by the mobile station 24 when communicating in the uplink direction on the RACH as directed by the network of the communication system 20. The shifted CAZAC sequence is used by the baseband processor for detecting a preamble of a communication received on the RACH from the mobile station 24 when the mobile station 24 is moving at a relatively high speed.
Shifting the CAZAC sequence essentially incorporates the frequency offset introduced by the mobile station motion into the reference sequence used for preamble detection. This approach utilizes information regarding additional peaks in a received communication, which are introduced because of the frequency offset, to collect back energy scattered (in the time domain) by the frequency offset.
A frequency offset that is sufficiently high results in additional correlation peaks at multiples of c_offoffsets, where c_off=(N_czm−1)/u for the u^thCazac root sequence, m is smallest positive integer for which c_offis integer and c_offcorresponds to the peak produced by a Doppler shift f_Dopp=1/T_PREand it depends on the root index u.
Assume that the transmitted RACH preamble is defined by the zero shift of the ZC sequence x_u(k). Then the samples of the received RACH preamble r(k), k=0, 1, . . . , N_zc−1, after the Doppler frequency shift of f_Dopp=1/T_PRE=1.25 KHz can be represented as
r(k)=x _u(k)e ^j2πf ^Dopp ^T ^sym ^k =x _u(k)W ^k, where W=e ^−j2π/N ^cz
where T_symis the duration of each RACH symbol in RACH preamble, T_sym=T_PRE/N_zc=0.95 us. Then,
$\begin{matrix} r (k) = W^{u [k (k + 1) + 2 (1 / u) k] / 2} \\ = W^{u (k^{2} + k + 2 c_{off} k + c_{off}^{2} + c_{off}) / 2} W^{- {uc}_{off} (c_{off} + 1) / 2} \\ = W^{u ⌊ (k (k + 1 + c_{off}) + c_{off} (k + c_{off} + 1) ⌋ / 2} W^{- {uc}_{off} (c_{off} + 1) / 2} \\ = W^{u (k + c_{off}) (k + c_{off} + 1) / 2} W^{- {uc}_{off} (c_{off} + 1) / 2} \\ = x_{u} (k + c_{off}) W^{- (u + 1) / 2 u}, \end{matrix}$
where, c_off=1/u, or equivalently c_off=(N·m−1)/u, for the smallest m such that c_offis an integer.
The last expression shows that the received RACH preamble after the Doppler frequency shift f_Dopp=1/T_PREis equal to the transmitted RACH preamble cyclically shifted by c_offwhere the transmitted RACH preamble is obtained from u^throot ZC sequence. The complex scaling constant in the received RACH preamble has unit magnitude and thus does not influence the correlation detector in the receiver.
An illustration of this effect is shown in FIG. 2. Additional peaks are illustrated at 30 for a given frequency offset. The peak without frequency offset would be at the vertical axis shown at 32. For a different frequency offset value the magnitude of the peaks at 30 will be different but the location of the additional peaks does not change. The locations of the additional peaks only depend on the root CAZAC sequence index u.
For a given frequency offset value, the large peaks concentrate in a limited number of locations. Therefore, for a given the CAZAC sequence index u and for a user whose nominal shift is d, one example considers the triple peaks at locations d and (d+/−c_off)mod Ncz, and uses proper combining to combat the scattering effect resulting from frequency offset.
The ZC sequence of odd length is given as
$a_{u} (k) = \exp (- jπ u \frac{k (k + 1)}{N_{ZC}})$
where u is the index of the root sequence, N_ZCis the length of the sequence, and k=0, 1, . . . N_ZC−1 is the index of the samples.
Assume that a_u,d(k)=a (k−d mod N_ZC) refers to the d^thcyclic shift of the root sequence u. k is the index in time that is of interest (i.e., in the timing uncertainty window) to the use with nominal shift d. The correlation values of the cyclic shift triplet can be described as {a_u,(d−c _off _{mod N} _G ₎(k), a_u,d(k), a_u,(d+c _off _{mod N} _G ₎(k)}. This triplet may be combined coherently or non-coherently to improve performance.
In one example that includes coherent combining, the frequency offset is estimated to calculate the coefficients. One example includes a hypothesis test similar to the technique described above. In one example, 3 or 5 hypotheses are considered, each corresponding to a possible frequency offset. For each uncertainty offset k for each hypothesis, a metrics can be done by calculating:
y _u,d,f(k)=a _u,(d−c _off _{mod N} _G ₎(k)b _u(−f)+a _u,d(k)+a _u,(d+c _off _{mod N} _G ₎(k)b _u(f)
where b_u(−f) and b_u(f) are predetermined coefficients based on the index u and an assumed frequency offset f. The value of b_u(f) (and similarly b_u(−f)) is given as:
$\begin{matrix} b_{u} (f) = \sum_{k = 0}^{Ncz - 1} a_{u} (k) \exp (- j2π {fkT}_{s}) \\ = \sum_{k = 0}^{Ncz - 1} a_{u} (k) \exp (- jπ (2 {fkT}_{s} + u \frac{k (k + 1)}{N_{cz}}) . \end{matrix}$
In one example that includes non-coherent combining, sorting and combining is done on a per user base. One example combines the energy coming from multiple receive antennas, multiple paths and shifted copies due to frequency offset. For a given user, the available correlation values are C_L ^m(n), where L is an antenna index, L=1 or 2, n is the offset index, n=1, . . . , 1.22 Nmp, and m is indicating one of the triplet elements resulting from the frequency offset, for example, m=−1, 0 and 1. In one example, m=0 means original offset and −1 and +1 indicates images that are −C_offand +C_offapart.
One example approach is summarized in the flow chart diagram 40 of FIG. 3. At 40, the power of the correlation values from each antenna, for each offset and each frequency offset window are calculated from
$p_{m} (n) = \sum_{L = 1}^{2} {\langle c_{L}^{m} (n) \rangle}^{2},$
for each offset n and each frequency offset triplet m.
All correlation powers higher than a threshold TH1 are determined at 44, for each triplet index m. Let us call the set of offsets Sm. S_m={δ_m(n)}, where n=1 . . . Nm, and Nm is the number of peaks found for each triplet index m.
At 46, the qualified powers within each frequency offset window are combined. The qualified power for a triplet index m can be represented as
$p_{m} = \sum_{n \in S_{m}} p_{m} (n)$
At 48 p_mis compared with another threshold TH2 to decide which frequency offset window to keep. One example includes keeping all p_mif it is greater than TH2 except for the case when both p−1 and p₁are greater than TH2 but p₀is less than Th2. In this case, the larger of P₋₁and p₁is kept. Let us call the set M.
The step schematically shown at 50 includes combining the p_maccording to the qualified frequency offset window
$p = \sum_{m \in M} p_{m}$
At 52 sorting and storing includes sorting and storing all combined power p that is higher than another threshold TH3 or sort and store N (e.g., N=8 or 16) highest power p.
This example includes combining the offset values corresponding to the strongest peaks in the qualified offset at 54. The way to combine can be, for example, by interpolation, possibly weighted.
$δ = \sum_{m \in M} \sum_{n = 1}^{N_{m}} δ_{m} (n) p_{m} (n) / \sum_{m \in M} p_{m}$
For simplicity, one example includes the time offset δ_mcorresponding to the highest peak in Sm, (i.e., δ_m=max(δ_m(n)), for n=−1. Nm), then,
$δ = \sum_{m \in M} δ_{m} p_{m} / \sum_{m \in M} p_{m}$
At 56, the combined power and the corresponding offset (p, δ) are reported to the peak search and report unit for further processing.
One example includes predetermining a set of shifted CAZAC sequences to be used for particular frequency offsets. The look up table is consulted based on an estimated frequency offset range in one example to select at least one shifted CAZAC sequence for detecting the preamble in a RACH communication.
One example includes attempting to use the root sequence and at least two shifted sequences, one on each side of the root sequence. The sequence that provides the highest peak is chosen for preamble detection.
The preceding description is exemplary rather than limiting in nature. Variations and modifications to the disclosed examples may become apparent to those skilled in the art that do not necessarily depart from the essence of this invention. The scope of legal protection given to this invention can only be determined by studying the following claims.

Claims

1. A method of communicating, comprising the steps of:

shifting a constant amplitude zero autocorrelation (CAZAC) root sequence used by a source of a received communication to a shifted CAZAC sequence; and

detecting a preamble in the received communication using the shifted CAZAC sequence.

2. The method of claim 1, comprising shifting the CAZAC root sequence by an amount corresponding to a length of the root sequence.

3. The method of claim 1 comprising

estimating a frequency offset of the received communication and selecting an amount of the shifting based on the estimated range of frequency offset.

4. The method of claim 1 comprising

performing the detecting using the root sequence, a first shifted sequence offset from the root sequence in a first direction and a second shifted sequence offset from the root sequence in a second, opposite direction; and

determining which of the root sequence, the first shifted sequence or the second shifted sequence detects a highest value of the received communication for frequency offset compensation caused by a Doppler effect.

5. The method of claim 1, comprising

predetermining a plurality of shifted CAZAC sequences, each corresponding to a root sequence and a frequency offset;

storing the predetermined plurality of shifted sequences in a look up table; and

selecting at least one of the shifted sequences from the look up table for performing the detecting based upon an expected range of frequency offset.

6. The method of claim 1, comprising

determining whether an effect of a frequency offset to the received communication exceeds a selected threshold; and

using the shifted sequence if the frequency offset exceeds the threshold and otherwise using the root sequence for the detecting.

7. The method of claim 1, wherein the received communication comprises a random access channel (RACH) communication.

8. A receiver device, comprising

a detector module configured to

shift a constant amplitude zero autocorrelation (CAZAC) root sequence used by a source of a received communication to a shifted CAZAC sequence and

detect a preamble in the received communication using the shifted CAZAC sequence.

9. The device of claim 8, wherein the detector module is configured to shift the CAZAC root sequence by an amount corresponding to a length of the root sequence.

10. The device of claim 8, wherein the detector module is configured to estimate a frequency offset of the received communication and selecting an amount of the shifting based on an estimated range of frequency offset.

11. The device of claim 8, wherein the detector module is configured to

use the root sequence, a first shifted sequence offset from the root sequence in a first direction and a second shifted sequence offset from the root sequence in a second, opposite direction to detect the preamble; and

determine which of the root sequence, the first shifted sequence or the second shifted sequence detects a highest value of the received communication for frequency offset compensation caused by a Doppler effect.

12. The device of claim 8, wherein the detector module is configured to

predetermine a plurality of shifted CAZAC sequences, each corresponding to a root sequence and a frequency offset;

store the predetermined plurality of shifted sequences in a look up table; and

select at least one of the shifted sequences from the look up table for performing the detecting based upon an expected range of frequency offset.

13. The device of claim 8, wherein the detector module is configured to

determine whether an effect of a frequency offset to the received communication exceeds a selected threshold; and

use the shifted sequence if the frequency offset exceeds the threshold and otherwise use the root sequence to detect the preamble.

14. The device of claim 8, wherein the received communication comprises a random access channel (RACH) communication.