US 6,961,364 B1Grant

Base station identification in orthogonal frequency division multiplexing based spread spectrum multiple access systems

Issue Date:2005-11-01

•52 Claims

•5 Drawing Sheets

Abstract

A base station having the strongest downlink signal is identified by utilizing a unique slope of a pilot tone hopping sequence being transmitted by a base station. Specifically, base station identification is realized by determining the slope of the strongest received pilot signal, i.e., the received pilot signal having the maximum energy. In an embodiment of the invention, the pilot tone hopping sequence is based on a Latin Squares sequence. With a Latin Squares based pilot tone hopping sequence, all a mobile user unit needs is to locate the frequency of the pilot tones at one time because the pilot tone locations at subsequent times can be determined from the slope of the Latin Squares pilot tone hopping sequence. The slope and initial frequency shift of the pilot tone hopping sequence with the strongest received power is determined by employing a unique maximum energy detector.

Metadata

Assignee

Flarion Technologies, Inc.

Inventors

Rajiv Laroia
Junyi Li
Sundeep Rangan
Pramod Viswanath

Application Information

Application Number:US 09/551,078

Filing Date:2000-04-18

Priority Date:2000-04-18

Art Unit:2631

Patent Drawings (5 sheets)

Description

Related Application

[0001] This application is related U.S. patent application Ser. No. 09/551,791 which was filed on Apr. 18, 2000.

Technical Field

[0002] This invention relates to wireless communications systems and, more particularly, to orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access (SSMA) systems.

Background of the Invention

[0003] It is important that wireless communications systems be such as to maximize the number of users that can be adequately served and to maximize data transmission rates, if data services are provided. Wireless communications systems are typically shared media systems, i.e., there is a fixed available bandwidth that is shared by all users of the wireless system. Such wireless communications systems are often implemented as so-called “cellular” communications systems, in which the territory being covered is divided into separate cells, and each cell is served by a base station.

[0004] In such systems, it is important that mobile user units are rapidly able to identify and synchronize to the downlink of a base station transmitting the strongest signal. Prior arrangements have transmitted training symbols periodically for mobile user units to detect and synchronize to the associated base station downlink. In such arrangements, there is a large probability that delays occur in identifying the base station transmitting the strongest signal because the training symbols are typically transmitted at the beginning of a frame. It is also likely that the training symbols transmitted from different base stations would interfere with each other. Indeed, it is known that once the training symbols interfere with each other they will continue to interfere. Thus, if the training symbols are corrupted, then the data is also corrupted, thereby causing loss in efficiency.

Summary of the Invention

[0005] Problems and/or limitations related to prior mobile user units that have attempted to identify a base station having the strongest downlink signal are addressed by utilizing a pilot tone hopping sequence being transmitted by a base station. Specifically, base station identification is realized by determining the slope of the strongest received pilot signal, i.e., the received pilot signal having the maximum energy.

[0006] In an embodiment of the invention, the pilot tone hopping sequence is based on a Latin Squares sequence. With a Latin Squares based pilot tone hopping sequence, all a mobile user unit needs is to locate the frequency of the pilot tones at one time because the pilot tone locations at subsequent times can be determined from the unique slope of the Latin Squares pilot tone hopping sequence. The slope and initial frequency shift of the pilot tone hopping sequence with the strongest received power is determined by employing a unique maximum energy detector. This unique slope of the pilot tone hopping sequence is then advantageously employed to identify the base station having the strongest downlink signal.

[0007] In one embodiment, the slope and initial frequency shift of the pilot signal having the strongest received power is determined by finding the slope and initial frequency shift of a predicted set of pilot tone locations having the maximum received energy.

[0008] In another embodiment, the frequency offset of the pilot signal with the strongest, i.e., maximum, received power is estimated at each of times “t”. These frequency offsets are employed in accordance with a prescribed relationship to determine the unknown slope and the initial frequency shift of the pilot signal.

[0009] A technical advantage to using the pilot tone hopping sequence to identify the base station having the strongest downlink signal is that the inherent latency resulting from using a sequence of training symbols is not present.

Brief Description of the Drawing

[0010] FIG. 1 illustrates a frequency domain representation in which a prescribed plurality of tones is generated in a prescribed bandwidth;

[0011] FIG. 2 illustrates a time domain representation of a tone ƒ_i;

[0012] FIG. 3 is a graphical representation of a time-frequency grid including a pilot tone hopping sequence;

[0013] FIG. 4 is a graphical representation of a Latin Squares hopping sequence;

[0014] FIG. 5 shows, in simplified block form, an OFDM-SSMA cellular system with Latin Squares pilots;

[0015] FIG. 6 shows, in simplified block diagram form, details of a mobile user unit in which an embodiment of the invention may advantageously be employed;

[0016] FIG. 7 shows, in simplified block diagram for, details of an embodiment of a maximum energy detector that may be employed in the mobile user unit of FIG. 6;

[0017] FIG. 8 shows, in simplified block diagram form, details of another embodiment of a maximum energy detector that may be employed in the mobile user unit of FIG. 6; and

[0018] FIG. 9 is a flow chart illustrating steps in a process that may be employed in the slope-shift solver of FIG. 8.

Detailed Description

[0019] FIG. 1 illustrates a frequency domain representation in which a prescribed plurality of tones is generated in a prescribed bandwidth. In this example, bandwidth W is employed to generate a total of N tones, i.e., i=1, . . . , N. The tones are spaced at Δƒ=1/T_Sapart, where T_sis the duration of an OFDM symbol. Note that the tones employed in this embodiment of the invention are generated differently than those generated for a narrow band system. Specifically, in a narrow band system the energy from each tone is strictly confined to a narrow bandwidth centered around the tone frequency, whereas in an Orthogonal Frequency Division Multiplexing (OFDM) system that is a wide band system the energy at a particular tone is allowed to leak into the entire bandwidth W, but it is so arranged that the tones do not interfere with one another.

[0020] FIG. 2 illustrates a time domain representation of tone ƒ_iwithin symbol interval T_s. Again, note that within each symbol interval T_s, data may be transmitted on each of the tones substantially simultaneously.

[0021] FIG. 3 is a graphical representation of an example OFDM channel from a base station to a number of mobile user units, i.e., receivers. The OFDM channel is represented as a time-frequency grid, i.e., plane. Each column of the grid represents the time interval for one OFDM symbol interval, and each OFDM symbol is comprised of a number of tones. In this example, there are N=5 tones in each symbol interval. The tones are numbered along the frequency axis and the symbol intervals, i.e., periods, are numbered along the time axis. If the spacing between tones in FIG. 3 is Δƒ, then:

- [0022] tone 0 corresponds to ƒ;
- [0023] tone 1 corresponds to ƒ+Δƒ;
- [0024] tone 2 corresponds to ƒ+2Δƒ;
- [0025] tone 3 corresponds to ƒ+3Δƒ;
- [0026] tone 4 corresponds to ƒ+4Δƒ.
  Similarly, if the duration of a symbol interval is T_sthen:
- [0027] time 0 corresponds to t₀;
- [0028] time 1 corresponds to t₀+T_s;
- [0029] time 2 corresponds to t₀+2 T_s;
- [0030] time 3 corresponds to t₀+3 T_s;
- [0031] time 4 corresponds to t₀+4 T_s;
- [0032] time 5 corresponds to t₀+5 T_s;
- [0033] time 6 corresponds to t₀+6 T_s.

[0034] In general, a pilot signal includes known waveforms that are transmitted from a base station so that mobile user units, i.e., receivers, can identify the base station and estimate various channel parameters. In an Orthogonal Frequency Division Multiplexing based Spread Spectrum Multiple Access (OFDM-SSMA) system, in accordance with an aspect of the invention, the pilot signal is comprised of known symbols transmitted on prescribed tones during prescribed symbol intervals. In a given symbol interval, the tones used for the pilot signal are called the “pilot tones”, and the assignment of pilot tones as a function of time is called the “pilot hopping sequence”. Again, it is noted that the inherent delays resulting when using the training sequence of symbols is not experienced when using the pilot tone hopping sequence to identify the base station having the strongest downlink signal.

[0035] Since the OFDM-SSMA physical layer is based on the pilot signals, symbols on the pilot tones are transmitted at higher power than symbols on non-pilot tones. Pilot tones are also boosted in power so that they may be received throughout the cell. Therefore, for the purpose of identification, pilot signals can be distinguished by the fact that the energy received on the pilot tones is higher than the energy on the non-pilot tones.

[0036] In FIG. 3, an example set of pilot tones is indicated by the hatched squares in the time-frequency grid. In this example, the base station transmits one pilot tone in each OFDM symbol interval. During: symbol interval (0), tone (1) is used as a pilot tone; symbol interval (1), tone (4) is used as a pilot tone; symbol interval (2), tone (0) is used as a pilot tone; symbol interval (3), tone (2) is used as a pilot tone; symbol interval (4), tone (4) is used as a pilot tone; symbol interval (5), tone (1) is used as a pilot tone; etc. . . .

[0037] FIG. 4 shows an example of a Latin Squares pilot hopping sequence. The pilot signal corresponding to a Latin Squares pilot hopping sequence will be called a “Latin Squares pilot signal”, or simply “Latin Squares pilot”. In a Latin Squares pilot hopping sequence, the number of tones, N, is a prime number, and the pilot signals are transmitted on a fixed number, N_p, of the N tones in each OFDM symbol interval. The tone number of the j-th pilot tone in the t-th symbol interval is given by,
σ_s(j, t)=st+n_j(mod N), j=1, . . . , N_p (1)
where s and n_jare integers. A Latin Squares pilot signal of the form of Equation (1) can be viewed as a set of N_pparallel, cyclically rotating lines in a prescribed time-frequency grid, i.e., plane. The parameter, s, is the slope of the lines and the parameters, n_j, are the frequency offsets. In the example Latin Squares pilot hopping in FIG. 4, N=11, N_p=2, n₁=1, n₂=5 and s=2.

[0038] The frequency offsets and slope are design parameters of the Latin Squares pilot signal. For the purpose of channel estimation, the frequency offsets and slope should be selected so that the pilot tones are close to uniformly distributed in the time-frequency plane. A uniform distribution minimizes the worst-case interpolation error in the channel estimation. Specific values for the frequency offsets and slopes can be tested by numerical simulation with a specific channel estimator and channel conditions.

[0039] FIG. 5 depicts an OFDM-SSMA cellular system using Latin Squares pilots. The figure shows two base stations 501 and 502 in the cellular system, denoted BS1 and BS2, respectively. Each base station 501, 502 in the cellular system transmits a Latin Squares pilot. A mobile user unit 503, denoted MS, receives pilots signals and other transmissions from one or more base stations in the cellular system. The Latin Squares pilots transmitted by all the base stations 501, 502 use the same total number of tones, N, number of pilot tones per OFDM symbol, N_p, and the frequency offsets, n_j. However, the slope, s, of each pilot signal is locally unique in the sense that no two neighboring base stations use the same slope. Each slope, s, is taken from some set Sε{0, 1, . . . , N−1}. The use of locally unique slopes minimizes collisions between pilot signals from neighboring base stations. In addition, the slope provides a unique identifier for each base station. In FIG. 6, the slope of the pilot signal from BS1 (501) is denoted s₁, and the slope of the pilot signal from BS2 (502) is denoted s₂.

[0040] The base station identification problem is for the mobile user unit 503 to estimate the slope, s∈S, of the strongest received pilot signal. To perform this identification, the mobile user unit 503 can be pre-programmed with the common pilot signal parameters, N, N_pand n_j, as well as the set of possible slopes, S.

[0041] In general, base station identification is conducted prior to downlink and carrier synchronization. Consequently, a mobile user unit 503 may receive the pilot signals with unknown frequency and timing errors, and mobile user units must be able to perform base station identification in the presence of these errors. Also, after identifying the pilot hopping sequence of the strongest base station, the mobile user unit must synchronize its timing and carrier so that the location of subsequent pilot tones can be determined.

[0042] To define this synchronization problem more precisely, let Δt denote the timing error between a base station and mobile user unit in number of OFDM symbol intervals, and Δn denote the frequency error in number of tones. For the time being, it is assumed that Δt and Δn are both integer errors. Fractional errors will be considered later. Under integer time and frequency errors, Δt and Δn, if a base station transmits a pilot sequence given by Equation (1), the j-th pilot tone in the t-th symbol interval of the mobile will appear on tone number,
σ_s(j,t+Δt)+Δn=b(t)+n_j, (2)
where,
b(t)=s(t+Δt)+Δn, (3)
and where b(t) is the pilot frequency shift at time t. Equation (2) shows that if the frequency shift b(t) is known, the locations of the pilot tones at t are known. Also, if the frequency shift is determined at any one time, say b(0), the frequency shift at other times can be determined from b(t)=b(0)+st. Therefore, for synchronization, it suffices to estimate the frequency shift at any one time. The value b(0) will be called the initial frequency shift.

[0043] The fact that synchronization requires only the estimation of the initial frequency shift is a particular and useful feature of the Latin Square pilot hopping sequences. In general, synchronization involves estimation of time and frequency errors, and therefore demands a two parameter search. Synchronization for the Latin Squares sequences considered here, however, only requires the estimation of one parameter.

[0044] In summary, in an OFDM-SSMA cellular system, each base station transmits a Latin Squares pilot signal with a locally unique slope. A mobile user unit performs base station identification by estimating the slope of the strongest received pilot signal. In addition, the mobile user unit can synchronize to the pilot signal by estimating its initial frequency shift.

[0045] FIG. 6 shows, in simplified block diagram form, the details of a mobile user unit 600 containing the proposed maximum energy detector for base station identification. An incoming signal is supplied via an antenna 601 to a down conversion unit 602. The incoming signal includes pilot signals from one or more base stations. Down conversion unit 602 yields the baseband signal r(t) from the signal received by the mobile user unit 600. The received signal r(t) is supplied to fast Fourier transform (FFT) unit 603 that during each OFDM symbol interval performs an FFT on it to yield Y(t,n). In this example, Y(t,n) denotes the complex value received on the n-th tone in the t-th symbol interval and is supplied to maximum energy detector 604 and to receiver 605. Maximum energy detector 604 uses FFT data Y(t,n) from N_syconsecutive OFDM symbols to estimate the slope and initial frequency shift of the pilot signal with the maximum received strength. As indicated above, the FFT symbols to be used for the base station identification are denoted Y(t,n), t=0, . . . , N_sy−1 and n=0, . . . , N−1, and the estimates of the slope and initial frequency shift of the strongest received pilot signal are denoted ŝ and {circumflex over (b)}₀,respectively. The pilot slope ŝ and initial frequency shift {circumflex over (b)}₀estimates are supplied to a receiver 605 and employed to synchronize receiver 605 to the incoming carrier and to locate subsequent symbols in the pilot signal.

[0046] FIG. 7 shows, in simplified block diagram form details of an embodiment of a maximum energy detector 604 that may be employed in the mobile user unit 600 of FIG. 6. It has been seen that for the Latin Squares pilot tones, each candidate slope, s, and initial frequency shift, b₀=b(0), corresponds to a set of predicted pilot tone locations, (t,n), with
n=st+b₀+n_j, j=1, . . . , N_p, t=0, . . . , N_sy−1. (4)
Symbols on these pilot tones should be received with greater power than the symbols on the non-pilot tones. That is, the energy, |Y(t,n)|², should on average be highest on the pilot tones of the pilot signal with the strongest received signal strength. Therefore, a natural way to estimate the slope and frequency shift of the strongest pilot signal is to find the slope and frequency shift for which there is a maximum received energy on the predicted set of pilot tone locations of Equation (4). The input to the maximum energy detector 604 of FIG. 6 is the FFT data, Y(t,n), t=0, . . . , N_sy−1 and n=0, . . . ,N−1. The slope-shift accumulator 701, accumulates the energy along each possible slope, s, and initial frequency shift, b₀. The accumulated energy is given by the signal:

\begin{matrix} J_{0} (s, b_{0}) = \sum_{t = 0}^{N_{sy} - 1} {\langle Y (t, st + b_{0} (\mod N)) \rangle}^{2} . & (5) \end{matrix}

Then, frequency shift accumulator 702 accumulates the energy along the pilot frequency shifts, namely:

\begin{matrix} J (s, b_{0}) = \sum_{j = 1}^{N_{p}} J_{0} (s, b_{0} + n_{j}) . & (6) \end{matrix}

Maximum detector 703 estimates the slope and frequency shift of the maximum energy pilot signal as the slope and frequency shifts corresponding to the maximum accumulated pilot energy, that is:

\begin{matrix} \begin{matrix} \hat{s}, {\hat{b}}_{0} = \arg \max_{s, b_{0}} J (s, b_{0}), \end{matrix} & (7) \end{matrix}

where the maximum is taken over sεS and b₀=0, . . . , N−1.

[0047] Unfortunately, in certain applications, the above computations of Equations (5), (6) and (7) may be difficult to perform in a reasonable amount of time with the processing power available at the mobile user unit 600. To see this, note that to compute J₀(s,b₀) in Equation (5) at a single point (s,b₀) requires N_syadditions. Therefore, to compute J₀(s,b₀) at all (s,b₀) requires NN_slN_syadditions, where N_slis the number of slopes in the slope set S. Similarly, computing J(s,b₀) in Equation (6) requires NN_slN_padditions. Therefore, the complete energy detector would require O(NN_sl(N_p+N_sy)) basic operations to perform. Therefore, for typical values such as N=400, N_sl=200, N_p=10 and N_sy=20, the full energy detector would require 2.4 million operations. This computation may be difficult for the mobile user unit 600 to perform in a suitable amount of time.

[0048] FIG. 8 shows, in simplified block diagram form details of another embodiment of a maximum energy detector that may be employed in the mobile user unit of FIG. 6. Symbolwise shift detector 801 estimates, at each time t, the frequency shift of the pilot signal with strongest received strength. Specifically, the block computes:

\begin{matrix} [E (t), n (t)] = \max_{n} \sum_{j = 1}^{N_{p}} {\langle Y (t, n + n_{j} (\mod N)) \rangle}^{2}, & (8) \end{matrix}

where E(t) is the maximum energy value and n(t) is the argument of the maximum. To understand the purpose of the computation in Equation (8), suppose that the tones of the strongest energy pilot signal appear at the locations, (t,n), given in Equation (4). Since the received energy |Y(t, n)|², will usually be maximum at these pilot tone locations, the maximization in Equation (9) will typically result in:
n(t)=st+b₀(mod N), (9)
and E(t) will typically be the pilot signal energy at the time t. The value n (t) in Equation (9) is precisely the frequency shift estimate of the pilot signal at time t. Note that n(t) is sometimes referred to as the symbolwise frequency shift estimate.

[0049] Slope-shift solver 802 uses the relation in Equation (9) and the frequency offset estimates, n(t), to determine the unknown slope, s, and initial frequency shift, b₀. Since, the pilot signals are only on average higher in power than the non-pilot tones, the relation of Equation (9) may not hold at all time points t. Therefore, the slope-shift solver 802 must be robust against some of the data points n(t) not satisfying Equation (9). For robustness, the value E(t) can be used as measure of the reliability of the data n(t). Larger values of E(t) imply a larger amount of energy captured at the frequency shift estimate, n(t), and such values of n(t) can therefore be considered more reliable.

[0050] One possible way of implementing a robust slope-shift solver 802 is referred to as the difference method. This method uses the fact that if n(t) and n(t−1) both satisfy Equation (10), then n(t)−n(t−1)=s. Therefore, the slope, s, can be estimated by:

\begin{matrix} \hat{s} = \arg \max_{s \in S} \sum_{t = 1}^{N_{sy} - 1} E (t) I_{{n (t) - n (t - 1) = s}} & (10) \end{matrix}

where 1 is the indicator function. The estimator as defined by Equation (10) finds the slope, s, on which the total received pilot energy, E(t), at the points, t, satisfying n(t)−n(t−1)=s is maximized. After estimating the slope, the initial frequency shift can be estimated by:

\begin{matrix} {\hat{b}}_{0} = \arg \max_{b_{0} = 0, \dots, N - 1} \sum_{t = 1}^{N_{sy} - 1} E (t) 1_{{n (t) = st + b_{0}}} . & (11) \end{matrix}

The difference method is the process given by Equations (10) and (11).

[0051] A second possible method for the slope-shift solver 802 is referred to as the iterative test method. FIG. 9 is a flow chart illustrating the steps for the iterative test solver

- [0052] Step 901: Start process.

[0053] Step 902: Initialize T={0, . . . , N_sy−1}, and E_max=0.

- [0054] Step 903: Compute $\begin{matrix} \begin{matrix} t_{0} = \arg \max_{t \in T} E (t) \\ [E_{0}, s_{0}] = \max_{s \in S} \sum_{t \in T} E (t) 1_{{n (t) = n (t_{0}) + s (t - t_{0})}} \\ T_{0} = {t \in T : n (t) = n (t_{0}) + s_{0} (t - t_{0})} \\ T = T \ T_{0} \end{matrix} & (12) \end{matrix}$
  where E₀is the value of the maximum, i.e., strongest value, and s₀is the argument of the maximum.
- [0055] Step 904: If E₀>E_max, go to step 905.
- [0056] Step 905: Set
  E_max=E₀,
  ŝ=S₀ (13)
  {circumflex over (b)}₀n=t₀)−S₀t₀.
  - [0057] Then, go to step 906.
- [0058] Step 904: If not go to step 906.
- [0059] Step 906: If T is non-empty return to step 903, otherwise END via step 907.
  The values ŝ and {circumflex over (b)}₀in Step 905 are the final estimates for the slope and initial frequency shift of the strongest pilot signal.

[0060] The logic in the iterative test method is as follows. The set T is a set of times and is initialized in Step 902 to all the N_sytime points. Step 903 then finds the time, t₀εT, and slope, s₀εS, such that the set of times t on the line n(t)=n(t₀)+s₀(t−t₀), has the largest total pilot signal energy. The points on this line are then removed from T. In Step 904, if the total energy on the candidate line is larger than any previous candidate line, the slope and frequency shift estimates are updated to the slope and frequency shifts of the candidate line in step 905. Steps 903 through 906 are repeated until all points have been used in a candidate line.

[0061] Both the difference method and iterative test method demand significantly less computational resources than the full maximum energy detector. In both methods, the bulk of the computation is in the initial symbolwise shift detection in Equation (8). It can be verified that to conduct this maximization at all the N_sytime points N_syNN_poperations. Therefore, for the values N=400, N_p=10 and N_sy=20, the simplified maximum energy detector would require 80000 operations, which is considerably less than the 2.4 million needed by the full energy detector.

[0062] The above base station identification methods can be further simplified by first quantizing the FFT data Y(t,n). For example, at each time t, we can compute a quantized value of Y(t,n) given by:

\begin{matrix} Y_{q} (t, n) = {\begin{matrix} 1 & if {\langle Y (t, n) \rangle}^{2} > q μ (t) \\ 0 & else \end{matrix} & (15) \end{matrix}

where q>1 is an adjustable quantization threshold, and μ(t) is the mean received energy at time t:

\begin{matrix} μ (t) = \frac{1}{N} \sum_{n = 0}^{N - 1} {\langle Y (t, n) \rangle}^{2} . & (16) \end{matrix}

The quantized value Y_q(t,n) can then be used in place of |Y(t,n)|²in the above base station identification processes. If the parameter q is set sufficiently high, Y_q(t,n) will be zero at most values of n, and therefore the computations such as Equation (8) will be simplified.

[0063] In the above discussion, it has been assumed that the time error between the base station and mobile is some integer number of OFDM symbol intervals, and the frequency error is some integer number of tones. However, in general both the time and frequency errors will have fractional components as well. Fractional errors result in the pilot tones being split between two time symbols and spread out in frequency. This splitting reduces the pilot power in the main time-frequency point, making the pilot more difficult to identify. Meanwhile, without proper downlink synchronization, data signals from the base station are not received orthogonally with the pilot signal, thus causing extra interference in addition to that generated by neighboring base-stations. Overall, fractional time and frequency errors can thereby significantly degrade the base station identification. In particular, the strongest energy detection process may not perform well.

[0064] To avoid this fractional problem, the above identification processes be run at several fractional offsets. Specifically, for a given received signal r(t), the mobile user unit can slide the FFT window N_ƒr,ttimes along the time axis, each time obtaining a different set of frequency sample vectors. The step size of sliding the FFT window should be 1/N_ƒr,tof the symbol interval. Similarly, the mobile user unit can slide the FFT window N_ƒr,ƒ times along the frequency axis with a spacing of 1/N_ƒr,ƒ of a tone. The identification process can be run on the frequency samples obtained from each of the fractional time and frequency offsets. This process yields N_ƒr,tN _ƒr,ƒcandidate slope and frequency shifts.

[0065] To determine which of the N_ƒr,tN _ƒr,ƒ candidate slope and shifts to use, the mobile user can select the slope and shift corresponding to the strongest pilot energy. For a given candidate (s,b₀) the pilot energy is given by J(s,b₀) in Equation (6). If the difference method is used, an approximation for the pilot energy is given by the value of the strongest attained in equation (11). The value E_maxmay be employed in the iterative test method.

[0066] The above-described embodiments are, of course, merely illustrative of the principles of the invention. Indeed, numerous other methods or apparatus may be devised by those skilled in the art without departing from the spirit and scope of the invention.

Claims

1. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system including at least two adjacent base stations, each one of the adjacent base stations transmitting pilot tones according to one of a plurality of different pilot tone hopping sequences over at least a portion of a pilot sequence transmission time period, said portion including multiple symbol time periods, at least one of the different pilot tone hopping sequences including at least two pilot tones per symbol time period which are separated from one another by at least one tone during said portion of said pilot sequence transmission time period, in each of the different pilot tone hopping sequences the number of pilot tones used in each successive symbol time periods in said portion of said pilot sequence transmission period being the same but the tones used in a symbol time period by any one of the different pilot tone hopping sequences changing in frequency from one symbol time period to the next symbol time period by a frequency shift corresponding to a fixed number of tones, adjacent base stations using different frequency shifts to generate pilot tone hopping sequences with different pilot tone slopes which can be determined from the frequency shift of the pilot tones used in consecutive symbol time periods, the apparatus comprising:

a receiver for receiving one or more of said plurality of different pilot tone hopping sequences having different pilot tone slopes; and

a detector, responsive to said one or more received pilot tone hopping sequences, said detector including an energy accumulator for generating an accumulated energy measurement for each individual one of the plurality of pilot tone hoping sequences having different slopes over a period including multiple symbol time periods, said detector detecting a received pilot tone hopping sequence having the maximum accumulated energy over said period including multiple symbol time periods.

2. The invention as defined in claim 1 wherein each of said one or more received pilot tone hopping sequences is a Latin Squares based pilot tone hopping sequence.

3. The invention as defined in claim 1 wherein said receiver yields a baseband version of a received signal and further includes a unit for generating a fast Fourier transform version of said baseband signal, and wherein said detector is supplied with said fast Fourier transform version of said baseband signal to detect, based on accumulated energy measurements, the received pilot tone sequence having the maximum accumulated energy.

4. The invention as defined in claim 3 wherein said receiver further includes a quantizer for quantizing the results of said fast Fourier transform.

5. The invention as defined in claim 3 wherein said detector is a maximum energy detector.

6. The invention as defined in claim 5, wherein different initial frequency shifts are possible for different pilot tone hopping sequences having the same slope; and wherein said maximum energy detector determines a slope and an initial frequency shift for pilot tones in the detected pilot tone hopping sequence having the maximum accumulated energy.

7. The method of claim 1, wherein frequency spacing between pilot tones which occur in a symbol time period in each of said plurality of tone hopping sequences is fixed and is the same for all of said plurality of pilot tone hopping sequences.

8. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising:

a receiver for receiving one or more pilot tone hopping sequences each including pilot tones, said pilot tones each being generated at a prescribed frequency and time instants in a prescribed time-frequency grid; and

a maximum energy detector, responsive to said one or more received pilot tone hopping sequences, for detecting the received pilot tone hopping sequence having the strongest power,

said maximum energy detector including a slope-shift accumulator for accumulating energy along each possible slope and initial frequency shift of said one or more received pilot tone hopping sequences and generating an accumulated energy signal, a frequency shift accumulator supplied with said accumulated energy signal for accumulating energy along pilot frequency shifts of said one or more received pilot tone hopping sequences, and a maximum detector supplied with an output from said frequency shift accumulator for estimating a slope and initial frequency shift of the strongest received pilot tone hopping sequence as a slope and initial frequency shift corresponding to the strongest accumulated energy.

9. The invention as defined in claim 8 wherein said accumulated energy is represented by the signal J₀(s, b₀), where

J_{0} (s, b_{0}) = \sum_{t = 0}^{N_{sy} - 1} {\langle Y (t, st + b_{0} (\mod N)) \rangle}^{2},

and s is the slope of the pilot signal, b₀is an initial frequency shift of the pilot signal, Y(t,n) is the fast Fourier transform data, t=0, . . . , N_sy−1, n=st+b₀(mod N), and n=0, . . . N−1.

10. The invention as defined in claim 8 wherein said frequency shift accumulator accumulates energy along pilot frequency shifts of said one or more received pilot tone hopping sequences in accordance with

J (s, b_{0}) = \sum_{j = 1}^{N_{p}} J_{0} (s, b_{0} + n_{j}),

where s is the slope of the pilot signal, b₀is an initial frequency shift of the pilot signal and n_jare frequency offsets.

11. The invention as defined in claim 8 wherein said maximum detector estimates said slope and initial frequency shift of the strongest received pilot tone hopping sequence in accordance with

\hat{s}, {\hat{b}}_{0} = \arg \max_{s, b_{0}} J (s, b_{0}),

where ŝ is the estimate of the slope, {circumflex over (b)}₀is the estimate of the initial frequency shift, and where the maximum is taken over sεS and b₀=0, . . . ,N−1.

12. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising:

a maximum energy detector, responsive to said one or more received pilot tone hopping sequences, for detecting the received pilot tone hopping sequence having the strongest power, said maximum energy detector including a frequency shift detector for estimating at a given time frequency shift of the received pilot tone hopping sequence having strongest energy and an estimated maximum energy value, and a slope and frequency shift solver, responsive to said estimated frequency shift and said estimated maximum energy value, for generating estimates of an estimated slope and an estimated initial frequency shift of the strongest received pilot signal.

13. The invention as defined in claim 12 wherein said estimated frequency shift at time t is obtained in accordance with n(t)=st+b₀(mod N), where s is the pilot signal slope, t is a symbol time and n(t) is a frequency shift estimate.

14. The invention as defined in claim 13 wherein said estimated maximum energy value is obtained in accordance with

[E (t), n (t)] = \max_{n} \sum_{j = 1}^{N_{p}} {\langle Y (t, n + n_{j} (\mod N)) \rangle}^{2},

where E(t) is the maximum energy value, Y(t,n) is the fast Fourier transform data, j=1, . . . , N_pand n_jare frequency offsets.

15. The invention as defined in claim 14 wherein said slope is estimated in accordance with

\hat{s} = \arg \max_{s \in S} \sum_{l = 1}^{N_{sy} - 1} E (t) 1_{{n (t) - n (t - 1) = s},}

where both n(t) and n(t−1) satisfy n(t)=st+b₀(mod N).

16. The invention as defined in claim 14 wherein said frequency shift is estimated in accordance with

b_{0} = \arg \max_{b_{0} = 0, \dots, N - 1} \sum_{t = 1}^{N_{sy} - 1} E (t) 1_{{n (t) = st + b_{0}} .}

17. The invention as defined in claim 12 wherein said maximum energy detector detects said slope in accordance with determining the time, t₀εT, and slope, s₀εS, such that the set of times t on the line n(t)=n(t₀)+s₀(t−t₀), has the largest total pilot signal energy.

18. A method for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system including at least two adjacent base stations, each one of the adjacent base stations transmitting pilot tones according to one of a plurality of different pilot tone hopping sequences, in each of the different pilot tone hopping sequences over at least a portion of a pilot sequence transmission time period, said portion including multiple symbol time periods, the number of pilot tones used in each successive symbol time period in said portion of said pilot sequence transmission time period being the same but the tones used in a symbol time period by any one of the different pilot tone hopping sequences changing in frequency from one symbol time period to the next symbol time period by a frequency shift corresponding to a fixed number of tones, adjacent base stations using different frequency shifts to generate pilot tone hoping sequences with different pilot tone slopes which can be determined from the frequency shift of the pilot tones used in consecutive symbol time periods, the method comprising the steps of:

receiving one or more of said plurality of different pilot tone hopping sequences having different pilot tone hopping slopes; and

in response to said one or more received pilot tone hopping sequences:

generating an accumulated energy measurement for each individual one of the plurality of pilot tone hoping sequences having different pilot tone hopping slopes over a period including multiple symbol time periods; and

detecting a received pilot tone hopping sequence having the maximum accumulated energy over said period including multiple symbol time periods.

19. The method as defined in claim 18 wherein each of said one or more received pilot tone hopping sequences is a Latin Squares based pilot tone hopping sequence.

20. The method as defined in claim 18 wherein said step of receiving yields a baseband version of a received signal and further including a step of generating a fast Fourier transform version of said baseband signal, and wherein said step of detecting is responsive to said fast Fourier transform version of said baseband signal for detecting the received pilot tone sequence having the maximum accumulated energy.

21. The method as defined in claim 20 wherein said step of receiving further includes a step of quantizing the results of said fast Fourier transform.

22. The method as defined in claim 20 wherein said step of detecting detects a maximum energy.

23. The method as defined in claim 22 wherein said step of detecting said maximum energy includes a step of determining a slope and initial frequency shift of pilot tones in a detected pilot tone hopping sequence having the maximum accumulated energy.

24. A method for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising the steps of:

receiving one or more pilot tone hopping sequences each including pilot tones, said pilot tones each being generated at a prescribed frequency and time instants in a prescribed time-frequency grid; and

in response to said one or more received pilot tone hopping sequences, detecting the received pilot tone hopping sequence having the maximum energy, said step of detecting said maximum energy including the steps of accumulating energy along each possible slope and initial frequency shift of said one or more received pilot tone hopping sequences and generating an accumulated energy signal, in response to said accumulated energy signal, accumulating energy along pilot frequency shifts of said one or more received pilot tone hopping sequences, and in response to an output from said step of frequency shift accumulating, estimating a slope and initial frequency shift of the strongest received pilot tone hopping sequence as a slope and initial frequency shift corresponding to the strongest accumulated energy.

25. The method as defined in claim 24 wherein said accumulated energy is represented by the signal J₀(s,b₀), where

J_{0} (s, b_{0}) = \sum_{t = 0}^{N_{sy} - 1} {\langle Y (t, st + b_{0} (\mod N)) \rangle}^{2},

26. The method as defined in claim 24 wherein said step of frequency shift accumulating includes a step of accumulating energy along pilot frequency shifts of said one or more received pilot tone hopping sequences in accordance with

J (s, b_{0}) = \sum_{j = 1}^{N_{p}} J_{0} (s, b_{0} + n_{j}),

where s is the slope of the pilot signal, b₀is an initial frequency shift of the pilot signal and n_jare frequency offsets.

27. The method as defined in claim 24 wherein said step of maximum energy detecting includes a step of estimating said slope and initial frequency shift of the strongest received pilot tone hopping sequence in accordance with

\hat{s}, {\hat{b}}_{0} = \arg \max_{s, b_{0}} J (s, b_{0}),

where ŝ is the estimate of the slope, {circumflex over (b)}₀is the estimate of the initial frequency shift, and where the maximum is taken over sεS and b₀=0, . . . ,N−1.

28. A method for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising the steps of:

in response to said one or more received pilot tone hopping sequences, detecting the received pilot tone hopping sequence having maximum energy, said step of detecting the received pilot tone hopping sequence having maximum energy including a step of estimating, at a given time, a frequency shift of the received pilot tone hopping sequence having maximum energy and estimating a maximum energy value, and in response to said estimated frequency shift and said estimated maximum energy value, generating estimates of an estimated slope and an estimated initial frequency shift of the strongest received pilot signal.

29. The method as defined in claim 28 wherein said estimated frequency shift at time t is obtained in accordance with n(t)=st+b₀(mod N), where s is the pilot signal slope, t is a symbol time and n(t) is a frequency shift estimate.

30. The method as defined in claim 29 wherein said estimated maximum energy value is obtained in accordance with

[E (t), n (t)] = \max_{n} \sum_{j = 1}^{N_{p}} {\langle Y (t, n + n_{j} (\mod N)) \rangle}^{2},

where E(t) is the maximum energy value, Y(t,n) is the fast Fourier transform data, j=1, . . . , N_pand n_jare frequency offsets.

31. The method as defined in claim 30 wherein said slope is estimated in accordance with

\hat{s} = \arg \max_{s \in S} \sum_{t = 1}^{N_{sy} - 1} E (t) 1_{{n (t) - n (t - 1) = s}},

where both n(t) and n(t−1) satisfy n(t)=st+b₀(mod N).

32. The method as defined in claim 30 wherein said frequency shift is estimated in accordance with

{\hat{b}}_{0} = \arg \max_{b_{0} = 0 \dots N - 1} \sum_{t = 0}^{N_{sy} - 1} E (t) 1_{{n (t) = st + b_{0}}} .

33. The method as defined in claim 28 wherein said step of maximum energy detecting includes a step of finding the time, t₀εT, and slope, s₀εS, such that the set of times t on the line n(t)=n(t₀)+s₀(t−t₀), has the largest total pilot signal energy.

34. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system including at least two adjacent base stations, each one of the adjacent base stations transmitting pilot tones according to one of a plurality of different pilot tone hopping sequences over at least a portion of a pilot sequence transmission time period, said portion including multiple symbol time periods, at least one of the different pilot tone hopping sequences including at least two pilot tones per symbol time period which are separated from one another by at least one tone during said portion of said pilot sequence transmission time period, in each of the different pilot tone hopping sequences the number of pilot tones used in each successive symbol time period in said portion of said pilot sequence transmission time period being the same but the tones used in a symbol time period by any one of the different pilot tone hopping sequences changing in frequency from one symbol time period to the next symbol time period by a frequency shift corresponding to a fixed number of tones, adjacent base stations using different frequency shifts to generate pilot tone hopping sequences with different pilot tone slopes which can be determined from the frequency shift of the pilot tones used in consecutive symbol time periods, the apparatus comprising:

means for receiving one or more of said different pilot tone hopping sequences each including pilot tones; and

means, responsive to said one or more received pilot tone hopping sequences, for generating an accumulated energy measurement for each individual one of the plurality of different pilot tone hoping sequences having different pilot tone slopes; and

detector means for detecting a received pilot tone hopping sequence having the maximum accumulated energy over a period including multiple symbol time periods.

35. The invention as defined in claim 34 wherein each of said one or more received pilot tone hopping sequences is a Latin Squares based pilot tone hopping sequence.

36. The invention as defined in claim 34 wherein said means for receiving yields a baseband version of a received signal and further including means for generating a fast Fourier transform version of said baseband signal, and wherein said means for detecting is responsive to said fast Fourier transform version of said baseband signal for determining a received pilot tone sequence having the maximum energy.

37. The invention as defined in claim 36 wherein said means for generating said fast Fourier transform includes means for quantizing the results of said fast Fourier transform.

38. The invention as defined in claim 36 wherein means for detecting detects a maximum energy.

39. The invention as defined in claim 38 wherein said means for detecting said maximum energy includes means for determining a slope and an initial frequency shift of pilot tones in a detected pilot tone hopping sequence having the maximum energy.

40. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising the steps of:

means for receiving one or more pilot tone hopping sequences each including pilot tones, said pilot tones each being generated at a prescribed frequency and time instants in a prescribed time-frequency grid; and

means, responsive to said one or more received pilot tone hopping sequences, for detecting the received pilot tone hopping sequence having maximum energy, said means for detecting said maximum energy including means for accumulating energy along each possible slope and initial frequency shift of said one or more received pilot tone hopping sequences, means for generating an accumulated energy signal, means, responsive to said accumulated energy signal, for accumulating energy along pilot frequency shifts of said one or more received pilot tone hopping sequences, and means, responsive to an output from said means for frequency shift accumulating, for estimating a slope and an initial frequency shift of the strongest received pilot tone hopping sequence as the slope and the initial frequency shift corresponding to the strongest accumulated energy.

41. The invention as defined in claim 40 wherein said accumulated energy is represented by the signal J₀(s₁b₀), where

J_{0} (s, b_{0}) = \sum_{t = 0}^{N_{sy} - 1} {\langle Y (t, st + b_{0} (\mod N)) \rangle}^{2},

and s is the slope of the pilot signal, b₀is an initial frequency shift of the pilot signal, Y(t,n) is the fast Fourier transform data, t=0, . . . N_sy−1, n=st+b₀(mod N), and n=0, . . . N−1.

42. The invention as defined in claim 40 wherein said means for frequency shift accumulating includes means for accumulating energy along pilot frequency shifts of said one or more received pilot tone hopping sequences in accordance with

J (s, b_{0}) = \sum_{j = 1}^{N_{p}} J_{0} (s, b_{0} + n_{j}),

where s is the slope of the pilot signal, b₀is an initial frequency shift of the pilot signal and n_jare frequency offsets.

43. The invention as defined in claim 40 wherein said means for maximum energy detecting includes means for estimating said slope and initial frequency shift of the strongest received pilot tone hopping sequence in accordance with

\hat{s}, {\hat{b}}_{0} = \arg \max_{s, b_{0}} J (s, b_{0}),

where ŝ is the estimate of the slope, {circumflex over (b)}₀is the estimate of the initial frequency shift, and where the maximum is taken over sεS and b₀=0, . . . ,N−1.

44. Apparatus for use in a mobile user unit in an orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising the steps of:

means, responsive to said one or more received pilot tone hopping sequences, for detecting the received pilot tone hopping sequence having maximum energy, said means for detecting said maximum energy including means for estimating at a given time a frequency shift of the received pilot tone hopping sequence having maximum energy and for estimating a maximum energy value, and means, responsive to said estimated frequency shift and said estimated maximum energy value, for generating estimates of an estimated slope and an estimated initial frequency shift of the strongest received pilot signal.

45. The invention as defined in claim 44 wherein said estimated frequency shift at time t is obtained in accordance with n(t)=st+b₀(mod N), where s is the pilot signal slope, t is a symbol time and n(t) is a frequency shift estimate.

46. The invention as defined in claim 45 wherein said estimated maximum energy value is obtained in accordance with

[E (t), n (t)] = \max_{n} \sum_{j = 1}^{N_{p}} {\langle Y (t, n + n_{j} (\mod N)) \rangle}^{2},

where E(t) is the maximum energy value, Y(t,n) is the fast Fourier transform data, j=1, . . . , N_pand n_jare frequency offsets.

47. The invention as defined in claim 46 wherein said slope is estimated in accordance with

\hat{s} = \arg \max_{s \in S} \sum_{t = 1}^{N_{sy} - 1} E (t) 1_{{n (t) - n (t - 1) = s}},

where both n(t) and n(t−1) satisfy

n(t)=st+b₀(mod N).

48. The invention as defined in claim 46 wherein said frequency shift is estimated in accordance with

{\hat{b}}_{0} = \arg \max_{b_{0} = 0 \dots N - 1} \sum_{t = 0}^{N_{sy} - 1} E (t) 1_{{n (t) = st + b_{0}}} .

49. The invention as defined in claim 44 wherein said means for detecting maximum energy includes means for finding the time, t₀εT, and slope, s₀εS, such that the set of times t on the line n(t)=n(t₀)+s₀(t−t₀), has the largest total pilot signal energy.

50. An orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless system comprising:

at least two adjacent base stations, each one of the adjacent base stations transmitting pilot tones according to one of a plurality of different pilot tone hopping sequences over at least a portion of a pilot sequence transmission time period, said portion including multiple symbol time periods, at least one of the different pilot tone hopping sequences including at least two pilot tones per symbol time period which are separated from one another by at least one tone during said portion of said pilot sequence transmission time period, in each of the different pilot tone hopping sequences the number of pilot tones used in each successive symbol time period in said portion of said pilot sequence transmission period being the same but the tones used in a symbol time period by any one of the different pilot tone hopping sequences changing in frequency from one symbol time period to the next symbol time period by a frequency shift corresponding to a fixed number of tones, adjacent base stations using different frequency shifts to generate pilot tone hopping sequences with different pilot tone slopes which can be determined from the frequency shift of the pilot tones used in consecutive symbol time periods; and

a mobile communications device including:

i) a receiver for receiving one or more of said plurality of different pilot tone hopping sequences; and

ii) means for determining the pilot tone slope of a received pilot tone hopping sequence.

51. An orthogonal frequency division multiplexing (OFDM) based spread spectrum multiple access wireless communications method, comprising:

at least two adjacent bases stations which transmit pilot tones according to different ones of a plurality of different pilot tone hopping sequences over at least a portion of a pilot sequence transmission time period, said portion including multiple symbol time periods, at least one of the different pilot tone hopping sequences including at least two pilot tones per symbol time period which are separated from one another by at least one tone during said portion of said pilot sequence transmission time period, in each of the different pilot tone hopping sequences the number of pilot tones used in each successive symbol time period in said portion of said pilot sequence transmission period being the same but the tones used in a symbol time period by any one of the different pilot tone hopping sequences changing in frequency from one symbol time period to the next symbol time period by a frequency shift corresponding to a fixed number of tones, each of the adjacent base stations using different frequency shifts to generate the transmitted pilot tone hopping sequences resulting in different pilot tone slopes which can be determined from the frequency shift of the pilot tones transmitted in consecutive symbol time periods.

52. The method of claim 51, wherein frequency spacing between pilot tones which occur in a symbol time period in each of said plurality of tone hopping sequences is fixed and is the same for all of said plurality of pilot tone hopping sequences.

Patent Citations (8)

Patent	Date	Inventor	Cited By
US5867478(A)	1999-02-01	Baum et al.	Examiner
US6018317(A)	2000-01-01	Dogan et al.	Examiner
US6131016(A)	2000-10-01	Greenstein et al.	Examiner
US6282185(B1)	2001-08-01	Hakkinen et al.	Examiner
US6473418(B1)	2002-10-01	Laroia et al.	Examiner
US2001/0043578(A1)	2001-11-01	Kumar et al.	Examiner
CA2298360	2000-09-01		Applicant
WO97/26742	1997-07-01		Applicant

Non-Patent Literature (7)

F. Tufvesson et al., “Pilot Assisted Estimation for OFDM in Mobile Cellular Systems”, IEEE, 0-7803-3659-3/97, pp. 1639-1643.Applicant
R. Negi et al., “Pilot Tone Selection for Channel Estimation in a Mobile OFDM System”, IEEE Transactions on Consumer Electronics, pp. 1122-1128, 1998.Applicant
EPC Search Report for European Application No. 01303316, Apr. 9, 2001.Applicant
Fernandez-Getino Garcia J et al.: “Efficient Pilot Patterns for Channel Estimation in OFDM Systems Over HF Channels”, VTC 1999-Fall, IEEE VTS 50th, Vehicular Technology Conference, Gateway to the 21st, Century Communications Village, Amsterdam, Sep. 19-22, 1999, IEEE Vehicular Technology Conference, New York, NY, U.S.A., vol. 4, Conf 50, Sep. 19, 1999, pp. 2193-2197.Applicant
Wang C C et al: “Dynamic Channel Resource Allocation in Frequency Hopped Wireless Communication Systems”, Information Theory, 1994. Proceedings, 1994 IEEE International Symposium on Trondheim, Norway 27, Jun. 1, Jul. 1994, New York, NY, U.S.A., IEEE, Jun. 27, 1994, p. 229.Applicant
Han D S et al: “On the Synchronization of MC-CDMA System for Indoor Wireless Communications”, VTC 1999-Fall, IEEE VTS 50th, Vehicular Technology Conference; Gateway to the 21stCentury Communications Village, Amsterdam, Sep. 19-22, 1999, IEEE Vehicular Technology Conference, New York, NY, U.S.A, vol. 2, Conf. 50, Sep. 1999, pp. 693-697.Applicant
Fazel K et al: “A Flexible and High Performance Cellular Mobile Communications System Based on Orthogonal Multi-Carrier SSMA”, Wireless Personal Communications, Kluwer Academic Publishers, NL, vol. 2, No. ½, 1995, pp. 121-144.Applicant

Abstract