GB2415872A - Varying pilot patterns in Multicarrier Spread Spectrum communication - Google Patents

Abstract

In adaptive Multi-Carrier (MC) Orthogonal Frequency Division Multiple Access (OFDMA) communication involving error correction coding, the pilot pattern and spreading factor are varied in order to adapt transmission performance. A denser pilot pattern, where the pilot signals are more closely spaced in time and frequency, models the channel more accurately but leaves less capacity for data. The error in the coded signal is calculated at the transmitting base station from the separation (in time or frequency) of each subcarrier from the closest pilot signal, using the indices and Signal to Noise Ratio (SNR) of the subchannels fed back from each mobile receiver and the calculated pairwise error probability (PEP) or Bit Eror Rate (BER) of a given code. The channel coding can be Turbo coding (PEP = equation 8) or convolutional coding (PEP = equation 13).

Description

MULT1CARRIER TRANSMISSION SYSTEMS This invention relates to a multi-

carrier transmission system m the context of wireless communications. More particularly, but not exclusively, the invention relates to channel coded multi-carrier systems, and methods of estimating the performance of such systems and allocating sub-carriers in such a system.

Backeround 1() Cellular services are now being used by millions of people every day and multi-media services are more and more in demand. The range of such multi-media services include for example short messaging, voice, data and video calls. Therefore, the required bit rate for the services vary widely from about I kbps for paging up to several Mbps for video transmissions. In order to support such a wide range of data rates and an even wider range for future systems, highly flexible mobility management is required and thus even more complex network structures are needed.

Multi-carrier (MC) systems are today commonly used and are a great success, with the IEEE standardization of for example IEEE 802.1 l a and g or IEEE 8()2.1 6a, or the ETSI stantlardisation of the Digital Audio Broadcasting (DAB) or the Digital Video Broadcasting (DVB-T) systems.

Multi-carricr modulation is a data transmission technique where several subcarrcrs are used to transport the user data signal.

Moreover, the enhancement of robustness to multi-path, flexibility and reduced complexity compared to other powerful techniques such as code division multiple access (CDMA) systems, has led to the introduction of new radio air interfaces, such as multicarrier CDMA systems (MC-COMA). Such systems are for example described in A.Chouly et al, "Orthogonal multicarrier techniques applied to direct sequence spread spectrum CDMA systems", GLOBECOM '93 or in N. Yee et al., "Multicarrier CDMA in Indoor Wireless Radio Networks", Proceedings PIMRC'93, pp. 126-133. MC systems have therefore been considered as a good candidate for Next Generation Wireless Enablers, see for example WWRF, Book of Vision, Rask 4.7, "MultiCarrier Based Air interface", pp232, http://www.wireless-world-research. org/, MAGN ET, http://www.telecom.ece.ntua.gr/magnet/ or WINNER, http://www.ist-winner.org/ Sixth Framework IST Project.

Next Generation air interface proposals involve thousands of subcarriers, and suggest user multiplexing by using Orthogonal Frequency Division Multiple Access systems (OFDMA), which has been proved to reduce intereell interference and sigmficant packet transmission flexibility.

S. Kaiser and K. Fazel proposed a hybrid scheme in the paper "A flexible Spread-Spectrum Multi-Carrier Multiple-Access System for Multi-Media Applications, m Proc. IEEE PIMRC'97, pp.l0()-104, namely MC-SS.

In MC-SS systems, frequency spreading of the user data is applied similarly to MC-CDMA systems. However, the orthogonality among the users is here obtained by multiplexing on distinct subcanriers. In this way the advantage of the frequency diversity offered by the spreading is combined with the flexibility of user multiplexing offered by OFDMA systems, see for example the paper by S. Pietrzyk and G.J.M Janssen, "Multiuser Subcarrier Allocation for QoS Provision in the OFDMA Systems", IEEE VTC 2002, Vol.2, pp 1 077- 1 08 1.

Transmission systems can be further improved by the use of channel coding. Channel coding is a method of adding redundancy to information so that it can be transmitted over a degrading (fading and/or noisy) channel, and subsequently be checked and corrected for errors that occurred in transmission. Channel coding is especially beneficial for wireless and multimedia applications. Channel coding is intimately connected with trellis modulation schemes such as quadrature amplitude modulation. The coding is characterized by the rate of the code, the level of the modulation and the code itself.

One example of a powerful coding scheme, the Turbo-Coding Scheme, was introduced by Bcrrou et al. in "Near Shannon limit error correcting coding and decoding: Turbo Codes", Proc. 1993, TEEE International Conference on Communications, May 1993, pp 1064-1070.

MC-SS transmission systems may be combined with Turbo-Codmg, for example in Turbo-Codng System using Diversity, as described in T. Ohtsuki, and J. M. Kahn, "Performance of Turbo Codes with Two-Branch Receive Diversity and Correlated Fading", VTC 2000 and A. Ramesh et al., "Performance Analysis of Turbo Codes on Fading Channels with Diversity Combining".

In future systems, such as for example 4G systems, the transmission performance is required to be varied on demand in order to provide a guaranteed 'Quality of Service' (QoS) to the users.

In the paper "Performance Analysis of Turbo Codes on Fading Channels with Diversity Combining" by A. Ramesh, a system is described to provide a way of adapting the transmission performance of MC-SS systems.

Multicarrier systems (such as Orthogonal Frequency Devision Multiple Access or OFDMA) rely on accurate phase and frequency synchronism in order to achieve high data rates. Accordingly, it is common to transmit pilot signals. These consist of sequences known to the receiving terminals, which are periodically sent from the base station. At the terminals, the pilot signals are compared with their expected values and the errors in frequency and phase are used to correct the local oscillator. Typically, pilot signals are sent on selected subcarriers, and the corrections for the intervening subcanriers (and intervening time periods) are calculated by interpolation.

The pilot signals therefore forth a pattern in time and frequency.

A more accurate result (and hence a lower error rate, and higher capacity) can be obtained by transmitting a denser pilot pattern -- i.e. spacing the pilot signals more closely in time and frequency. However, this reduces the amount of capacity available for data.

An object of the invention is therefore to provide a method of data transmission in a multicarrier system using error-correcting codes, in which the density of the pilot pattern is controlled to achieve a good balance of quality (i.e. error rate) against capacity.

We have realised that the parameters of the coding are also connected, since errors in frequency and phase can to some extent be compensated by the coding. We have established a relationship between the reception errors and the bit error rate experienced by a coding scheme, through calculation of the pairwise error probability. We therefore provide, in one aspect, a transmission system of this type in which the density of the pilot pattern (in time and/or frequency) is set as a function of the measured reception quality.

This is achieved by using a relationship which relates the separation (in frequency and/or time) of each subcarrier from the closest pilot signal to the maximum error this produces in the coded signal.

In it, an analysis has been carried out to provide performance evaluation for Turbo-Codes used in diversity systems, using the traditional Union upper bound framework. It is noted that in this analysis it is assumed that the weight factors involved in the diversity reception schemes are perfectly known andthat the channel estimates are perfectly known.

lt is therefore an aim of thc present invention to improve thc system described in the paper above and to provide an alternative and improved way of adapting the transmission performance of a channel coded system.

It is a further aim of the present invention to provide an improved estimate of the transmission performance of channel coded systems.

According to a further aspect of the present invention, there is provided a method of controlling a coded multicarricr telecommunications system which includes use of a measure of the quality of service (for example, BER) which includes a model of the errors on individual subcarriers. Alternatively, there is provided a method of controlling a multicarrier telecommunications system which includes use of a measure of the pairwise error probability of a code which includes a model of the errors on individual subcarriers. In either case, the model preferably calculates the subcarrier imperfections based on their spacing from a pilot 1 5 signal.

The measure may be used to design the codes used in the system, to allocate initial session parameters to a communications session with a tcnninal, or to adapt the session parameters during a communications session with a tenminal.ln this way, a more 'realistic' estimate of the transmission perf'ormancc, taking into account imperfect knowledge of the transmission system, is used to adapt the link efficiency for one particular data transmission. Thus the transmission perfonmancc can be adapted to the requirements of a user or a system in teens of the quality of service.

By taking into account both the limited knowledge about the channel model, for example by including maximal ratio combining and a measure for the imperfection of the channel estimates, meaningful adaptation schemes can be determined.

Preferably, the method is applied to turbo-coded multi-carrier spread spectrum systems.

In this way the method may be further improved by using a joint adaptation strategy based on adapting both the spreading factor and the pilot pattern. In this way the transmission performance can be adapted more efficiently to improve the capacity of the communications link. For example, if the use of extra pilot signals would be desirable to reduce the error rate, but this would result in an unacceptable loss of channel capacity, then the length of the spreading factor (i.e. the number of subcarriers over which the signal is spread) can be increased instead, to keep the capacity and the quality within 1 5 bounds.

Other aspects and preferred embodiments are disclosed in the claims and described hereinafter. At this point it may be mentioned that some previous background work has been proposed by the present inventor T. Lestable "Design of Link Adaptation Mechanisms for Future Generation Multi-Carrier Spread-Spectnm (MC-SS) Systems", PhD Thesis (in French), October 2003, Orsay University, Supelec, Paris, France, but this earlier work does not deal with the effects of coding, which are l;ndamental to the present invention.

These and other aspects, preferred features and embodiments will be described further, by way of example only, with reference to the accompanying drawings in which: 5Figure I is a block diagram illustrating a transmitter according to one embodiment of the present invention (for example, at a base station); Figure 2 is a diagram schematically illustrating a coder for Parallel Concatenated Convolutional Coding (Turbo Coding) at the transmitter; Figure 3 is a diagram schematically illustrating a decoder for iterative 10Turbo-dccoding at a receiver; Figured is a schematic graph illustrating the time-frequency pilot pattern of a multi-carrier system of an embodiment of the invention; Figure S is a graph illustrating the impact of channel delay spread on frequency correlation with respect to sub- carrier indices in a the embodiment; 15Figure 6 is a schematic diagram illustrating the dependence of the bit error rate on the distance from pilot tones for three different spreading factors; Figure 7 is a schematic graph illustrating qualitatively the pilot pattern in dependence on the STIR for different BER's; Figure 8 is a schematic diagram illustrating qualitatively the overhead 20as a function of the signal to noise ratio for different spreading factors; and Figure 9 is a schematic diagram illustrating qualitatively the joint adaptation of pilot pattern and spreading factors in dependence on the signal to noise ratio for dillercnt bit error rates; Figure 10 is a block diagram showing schematically the elements of a communications system according to an embodiment of the invention; Figure 11 is a block diagram showing schematically the elements of a base station or Node B of the network of Figure 10; Figure 12 is a corresponding block diagram showing schematically the structure of one of a plurality of terminals of that network; and Figure 13 is a schematic graph showing frequency on one horizontal axis and amplitude on the vertical axis, with several values of the (log) number of subcarriers used (i.e. spreading factor length) illustrating qualitatively the pilot pattern; Overview of first embodiment Referring to Figure 10, a communication system such as a mobile telephony system comprises a plurality of user equipment (UK) such as mobile terminals 300a, 300b, in radio communication with a base station 100, provided within a cell, and having a fixed link connection to a backbone network (such as an IP network) via a switch computer 200. The IP backbone network will not further be discussed since it is of conventional type.

The orthogonal frequency division multiple (OFDM) waveform used in the downlmk communications comprises a plurality of subcarriers spaced apart along the frequency axis such that the peak of each subcarrier falls withm the first null of its neighbour.

Referring to Figure 11, the base station 100 comprises a receiver antenna 102 coupled to an uplink receiver 104 which receives and demodulates uplink signals from the terminals 300. The received signal from the receiver 104 is demultiplexed by a demultiplexer 110 into data signals from each user terminal (which are routed to the IF network computer 200) and uplink control signals, which are supplied to a control unit 120.

On the transmitter side, data signals intended for terminals 300 are received from the network computer 200 and passed to a multiplexer 116 where they are multiplexed, together with control signals from the control unit 120, onto selected OFDM subcarriers generated by an OFDM generator 115.

The OFDM signal is then amplified and transmitted by a downlink transmitter 114 coupled to a downlink antenna 112, for reception by the mobile terminals 300.

A carrier allocator 115 determines which subcarriers will be used to communicate with each mobile terminal 300, on the basis of the uplink feedback signals from the mobile terminals, and controls the multiplexer accordingly.

Conveniently, the uplink transmissions are also OFDMA, but it would be possible to use any other multiple access system such as CDMA, TDMA, or random access protocols such as ALOHA.

Referring to Figure 12, the structure of a terminal transceiver 300 of Figure 10 is shown. A downlink receiver antenna 312 receives the OFDM signal transmitted by the base station 100 and passes it to a downlink receiver -1 1 314 at which it is amplified and down converted and supplied to an OFDM demodulator 315. Selected channels from the OFDM demodulator 315, under control of a channel selector 310, pass to a demultiplexer 316 which separates out the control signals intended to control the uplink from data signals. The download data signals are then combined using, for example, maximum ratio combining (or any other suitable form of combination) and supplied for use, at a data port (for example connected to a display unit, or an analog to digital converter for audio reproduction). Pilot signals are routed to a control unit 320.

Output signals from the uplink feedback control unit are supplied to a multiplexer 306, together with uplink data (for example, user commands or voice data), and the multiplexed uplink signal is supplied to an uplink transmitter 304 for transmission via an uplink antenna 302 to the base station (it will be appreciated that a separate uplink feedback channel could be provided, in which case the demultiplexer would be redundant).

It will be clear to the skilled person that, other than the modulation and RF components, the blocks making up the embodiment at the base station and the mobile terminals comprise suitably programmed digital signal processor devices (DSPs) or ASICs executing signal processing. Separate dedicated hardware devices may be used for the OFDM operations (specifically the Inverse Fast Fourier Transform or IFFT used in the transmitter components to map the subcarrers into a composite tme-domain signal for RF modulation) and the Fast Fourier Transform or FFT used in the receiver components to map the time-domain signal back into component subcarriers).

The operation of the system will now be described in overview.

Periodically (for example, every frame of a framed communication system) the base station 100 receives uplink feedback signals from the mobile terminals 300. Each uplink feedback signal comprises data indicating the signal to noise ratio (SNR) experienced on that subearrier.

The base station determines whether the current pilot pattern is sufficient to satisfy a maximum bit error rate criterion. If not, then the base station tests the effect of adding further pilot signals. If this would satisfy the maximum bit error rate criterion without reducing the capacity past a predetermined point, then the new pilot pattern is selected, and the base station signals to the terminal to change to the new pattern. If not, then the base station increases the spreading factor (i.e. the number of channels used to communicate with the terminal), and the base station signals to the terminal to change to the new spreading factor.

In order to transmit data over a communications link, data are digitally encoded, are modulated from digital to analog and amplified for transmission over the communications link.

In order to receive data transmitted over the communications hnk, the received data are first demodulated, then processed in error control decoders using the redundancy to correct any errors occurred in transmission, such that the data are then available at the receiver's output port (for example for display or audio reproduction).

Transmission performance is usually determined by measuring the Bit Error Rate (BER) and the Signal-to-Noise Ratio (SNR) of the transmitted signal at the reciever.

Figure 1 illustrates a block diagram of a transmitter 10 according to an embodiment of the present invention.

The binary data are encoded in channel coder 12, by coding the data bits, interleaving and mapping the coded data into constellation symbols.

The data stream is subsequently fed to a serial to parallel converter block 14 and is thereby multiplexed into N parallel data streams, i.e. the N Spreading Bands, constituted of L subcarriers each.

In block 16 each data stream is spread over Lsubcarriers using spreading chip codes Cal to CL, i.e. by duplicating the data stream L times and multiplying by the chip elements Car to CL (one chosen spreading sequence) Specific Frequency mapping is applied in block 18 and pilot tones (i.c.

predetermined symbols known to the receiver and carrying no data) arc introduced as will be described in more detail below. In block 20 an inverse Fast Fourier transformation is carried out and the data are fed into parallel to serial converter, wherein the data are added and sent through the channel for transmission.

Turbo-Codin & Decodina: An encoder for the Turbo encoding process (at the base station) is shownin Figure 2. This corresponds to the parallel concatenation of two recursive systematic convolutional (RSC) encoders. These encoders are separated by one interleaving module IT. As these encoders are processing the same mput bits stream xs, this scheme is designated as parallel concatenated convolutional codes (PCCC).

Each RSC component (here chosen to be identical) is composed of two generator polynomial, namely feedback and feedforward polynomials, respectively g, (D) and go (D). Thus, two parity check sequences are generated, namely x,r, and xP In order to efficiently vary the transmission rate, generally some puncturing can be processed on the parity check sequences.

The turbo-decoding process which takes places in the recievers at the base station and the terminals, performed by the decoder shown in Figure 3, is one of the most basic iterative scheme. Again two component decoders, called Soft Input Soft Output (SISO) decoders Do and D2, are serially concatenated via an interleaving process. The major key feature of this iterative process, is that the extrinsic information delivered by one SISO dccodcr, feeds the 2() following SlSO decoder as a priori information for decoding information bits.

The feedback loop tcrativcly improves the perfonnance of this scheme.

First Embodiment: Turbo-Coded Multi-Carrier Spread-Spectrum Realistic Channel Estimates Systems When designing Channel Codes, a realistic estimate of their performance is useful in order to evaluate the Coding for their future use.

As described above, in the paper by A. Ramesh, "Performance Analysis of Turbo Codes on Fading Channels with Diversity Combining", one first analysis has been carried out to provide performance evaluation for TurboCodes used in the environment of MC-SS transmission systems, using the traditional Union upper bound framework. It is noted that in this analysis it is assumed that the weight factors involved in the diversity reception schemes are perfectly known. This hypothesis simplifies the analysis, leading to optimal performance assessment. However, on the other hand, it is only a coarse approximation and does not allow one to consider realistic estimates for system design parameters.

In the following, by way of contrast, there is provided a detailed, theoretical analysis of the sensitivity of Turbo-Codes performance in MCSS, including the effects of Maximal Ratio Combining (MRC) and imperfect Channel Estimates, which will be used for selecting and adapting the pilot pattern and spreading factor.

In this way the upper bound performance can be estimated and used in such "degraded", and thus more realistic environments.

Traditional Union Bound Traditional Union upper bound for the Maximum Likelihood (ML) decoding of an (N,K) block code is given by: \ ) 1()' ]2(fl) 1 with A(d) being the number of codewords with Hamming weight do, and P2(d) is the probability of incorrectly decoding to a codeword with weight d.

In S. Benedetto, and G. Montorsi, "Unveiling Turbo Codes: Some Results on Parallel Concatenated Coding Schcmcs", by IEEE Trans. On Info.

Theory, Vol.42, p.409-429. March 1996, and D. Divsalar, S. Dolinar, R.J.

McEliece, and F. Pollara, Transfer Function Bounds on the Performance of Turbo Codes, TDA Progress Report 42-122, JPL, Caltech, August 1995, the authors propose an average upper bound averaged over all possible interleavers. Following their proposed framework, the average weight distribution is given by: 3= X| | r(lli) A! 1 = where i!(& ;' is the number of input words with Hamming weight I, and p(dl) is the probability that an input word with Hamming weight produces a codeword with Hamming wegllt d.

So, we act the average upper bound: - z --t=" V K 14 = t /li)-l' ('/) ct_t,.,, _ I _.( ). In' I t1' 1p and for the bit error probability: I/ AJ I'[i2(tl) 11 (1) Equation (l) is known as the Divsalar equation. It provides for calculation of the error probability (the BER and hence Quality of Service) given the pairwise error probability.

The key point in this embodiment is therefore to evaluate the pairwise probability of error in the context of a combining diversity scheme. In the following, the maximal ratio combination will be considered.

General Diversity Framework Let us note the output SNR from the receiving device: z(] )2 (2) = n (T2 1 2 YL/ then conditioned to the channel facings (sub-carriers), we get: p,\IR{()= /ll' }(0 -> -X,|) -3() as tt v2 rim, J are i i d, we can note f JIJ|; n n I r2 l, n (3) PI i' n-l all With Craig's formula ()(r)--- c.xl:' pinup: cad We finally get I. / fin ' {l 2) my| J Al (' ', 3 l: Sill- (' ) I o ' <1) = j J I',- (A) ' ' air] aid r 7 If we note /' .= J/.(',).,, 2.2V , then the resulting SNR {I distribution will impact the average pairwise error probability.

Turbo-Codin Upper Bound, with MRC and Imperfect Channel Estimate We will now introduce the Imperfect Channel Estimate by means of correlation factors between the pilot channel estimates and the true channel values.

The starting point Is thus the following double ntcgral: { ('l)=-:7Pr(r) 2 0 dr1 dg (4) We just have to focus on the derivation of this first integral: It= |pr)e29U''dY Based on the work by M.J. Cans, "The Effect of Gaussian Error in Maximal Ratio Combiners", IEEE Trans. On Comm. Tech., vol com-l9, N4, August 1971, the SNR distribution with Imperfect Channel Estimation is given by the following relation: Pr() =AJ(L p) (7'f ' rim (S) where by definition ^(k'.)-(', I) (I-p) p2( and p is the correlation factor between the true and estimated channel weight.

Note that Ark.= Bk l,(P) are the Bernstein Polynomials So, we get ,=(k.p) racy i- [7, expand ( rag a;,, 0)] r: (6) Now, we note for simplicity (I 2 Sin29)) dy Q=l+ I 2.sin and then by using the Gamma Euler function definition: (x)= its-' c-' , o We obtain straightforwardly Zk = r2k So by using this and equation (6), we now have: I,=,4(k,p) (r O)l 1 Sin2 = with Q since+ - and finally 1, =A(k p) singly I= Lsin2 +2] Thus, by using equation (3), P2 (d)= - - i1,) do That leads to the following relation: pA/RC' 1 sin 2 o 2 ( ) | EA(kp) [; 2 rig ]2 =-(k- p) r l 1 I s Byloti'g Hi= Lsin20+25 J weSiliplifythisexpression Indeed, consider the follo\\:ing notation: X(k p)= T'(k p) | sine 1 Sit12 0+ .! 1 We thus has c to focus on the new relation: 2 ( " ) For sake of simplicity, we note X(k-P)= Xk' This second integral is the expression of multinomial series, meaning that we can write 12 in the following: 12 = (Z PA) = !' !i]! it I: [+;.+ +;/=J With A,; rl nt sin 1 (I [A(k)]'^ I=! k=! LSln 0+ :] A=! 10then first compute, 6 sin l sin2 l(7, i: =,Lsin2t}+] =Lsin20+] also we get [ (k, p)] (k 1) (1-p p ^ for sake of clarity we note (i'i2'---;I''L'd'P)= j!; ! j! n[A(k,P)] iri ' n(k-I) (1_p2-ki5,p ZI( ); 1 5 1- 2---J/ - k=l Finally, this second integral can be rewritten as: _ k q I, 2 --y/(ii,i2, qii L d p) Sin20 il. ;, Sjn2O+I +'2+ t,/=J _ 2 By using this equation and equation (7), we obtain: + ,'2 ', (L,,p)=- Jl2 dd o ) ---Y/(il'i2 --,il,L,(l, p) I. I sin 1 dO il+;+ 4;,'1 T o sin20+r] And as Alouini uses the following integral family: Jn, (c)=-. | s 2 | do where parameters are given by c=E m=k /l =, As m is an integer, we can furthermore write: Jm(c)= [P(c)] ( k) [I P(C)] =0 witl1 P(C)=2 t1-: As a conclusion, thc ncw average pairwise error probablity. vith ] 5 1\1RC and nnpcrfcct clannel cstinatc is given hy ( 8) Thus, we have a new, more accurate measure of the pairuise error probability, which can be inserted into Equation (1) above to yield the BER.

The equation takes as inputs the average STIR reported by the terminal for the subcarriers, and the index of the subcarrier (W]liC]1 reneCtS]1OW far away the subcarrier is in frequency from the pilot tone). It is a function of L, the length of the spreading sequence (i.c. the number of subncarriers over which the signal to the terminal has been spread), d (which is the Hamming distance between codewords, a function of the code and level of modulation used) and p, the estimated error due to the subcarrier being spaced from the pilot. The BER can therefore be calculated for any terminal, and if it is too high it can be reduced by altering one of these variables.

From equation (8), it follows that the ideal case, i.e. with perfect channel estimate (p = I)is given by: 1 5 p\1Rt (L p = 1) = .1,; (-) Geometric Interpretation of the Pair,ise Error Probability We will demonstrate in this part, that based on the exprcsson of Eq(8), the Pairwise Error probability Uit]1 Ilamn1ing distance d of Turbo Coded transmission systems with L diversity MRC reception, and imperfect chancel estimate is actually the baryccntre of a n1csh nctworl: ubose point controls are the pairwise error prohability with Hamming distance varying from O to d and MRC reception with varying diversity from I to L, Hun the channel is perfectly known.

As described above, the weighting factor from Eq (8) was given by: Y'(/, '/2,- ./, .L,,I. p)= j tj, , n [A(k. p)] I' 2-' J.' I=' i,!i:!. , ! k=l (k-1) In order to demonstrate the geometric theorem, we need to estimatethe global summation of the above function with respect to all indices: mu.. y/(i'.i27,i/,L,d, p) [I 1. [L ,, ! i2! Hi, ! n ( k - 1) ( P I,,j, ,,nt(k_1) (l_p2Il-k);l (p21k-)tll it ': '/ 1' 2'''JL' k=l '2 ', i;!i2!. i, ! nt(k- 1) (] -P (P2k-l)] and now with A(k, p)=(k) (1 _ p2) p2 (k-1), and A(k, p) = Bk-,,-,(P), 7 Y/(i,,i2. ,i L,d. p) . i!i! i! [A(k. p)] [ A (k; p)] But due to the Bernstein polynomial property: A(k,p)=I k=l Indeed, (k,p)=(k_I) (1-X) X,W'th X= p so by noting k:=k-l, and N=L-I, this gives A(k, A) (K) (1 X) x = [(I-x)+x] = I k=l K=0 Finally we get the following remarkable property: (i,,i2, ,i,,L,d,p)=1 (9) 11 1, l/ Application to Time-Frequencv Pilot Pattern In order to illustrate the applicability of the estimate described above, an expression for the correlation factor p is derived as an example.

Fig. 4 illustrates a Time-Frequency Pilot Pattern, as is usually considered in multi-carrier based Systems.

In a first step, factors are analysed along each dimension, and subsequently, the 2D (time and frequency) correlation factor is derived Frequencv Correlation For an exponential decaying power delay profile, we can express the frequency correlation as: 41 + (2 Af rents)2 where p(Af) is the frequency separation between the pilot sub-carrier and the data sub-carrier, and rRi'S represents the Channel Delay Spread.

As the transmission system is MC based, we can note: Ts Ntuhc Nix the constant frequency width of sub-carriers.

Then introducing this parameter we get: Pk p(k fs'h) 2 (10) Jl + (2 k 6.f. ,.hC rains) where k is the sub-carrier index.

Time Correlation In the same manner bale can get the time correlation relation: p(^t) = Jo (2 Off, At) where At is the time separation between the pilot symbols and the data ones.

By introducing OFDM based paranletcrs such as

T I = _

Tl = To + Tg = (Nix, +Ng) To we can express the time correlation with the nth OFDM symbol: An =P(n T.)= pL[n (Off, +Ng) To]= Jo 27r n (Nix, +Ng) fit' (11) Figure 5 is a diagrram illustration of the impact of the channel delay spread on the frequency correlation with respect to the sub- can-ier index. Two examples are illustrated for a delay spread of 50 and 150 ns respectively.

Time-Frequency Correlation For the Time-Frequency case, we just need to combine both results about the correlation to get the expression of the 2D correlation factor: p(t,f)=P() p(k ^fuh)= tJo(2 I,' Ad) 41 + (2 k J;.,h ' r/JS) (12) Adaptive Pilot Pattern Guaranteeing QoS In the following a method is described in which the transmission performance of a communications system can be varied on demand, so as to fulfill the user's requirements in tempts of 'Qualify of Service'. (QoS) According to one enbodimcnt, this is achieved by varying the pilot pattcn1 in the timefrequency plane. Such a pilot pattcn1 Is illustrated in Fig. 4.

Tlowever, in order to link the transmission performance to the user's requirements of QoS, the relationship between the pilot pattern and the QoS requirement is required.

Such a relationship expressing the QoS in teems of the bit enror rate (BER), is given by equation (8) By applying the estimate of the BER upper bound of turbo-coded MC SS systems, considering "realistic" scenarios including MRC and imperfect channel estimates, the QoS can be expressed as a function of the pilot pattern.

Then, equation (12) replaces the perfect channel estimate (p=1), wherein the average pair wise error probability is given by 117 ( 1 (. it? = 1) --a I! A! ) Where Jm( ,+ I At| bill i' 1 ia and parameters are given by f r i c=_ i

L

17) - k it (=1 As m is an integer, we can furthermore write: -- [I,- )17, >:1 [I f'()lL [-_3 with Am)= 1. 1 It can be seen from equation (8), that for a given spreading factor (SF=L), the BER upper bound depends only on the correlation factor p. That means that by varying a single parameter (i.e. the spreading factor) the transmission performance can be varied or adapted to the user's requirements.

Pilot Pattern Selection The sub-carrier allocation principle is depicted in Fig. 6 below. Using equation (2), the error rate per sub-carrier (crosses) can be indeed evaluated, according to its frequency distance (10) from the pilot symbols.

For a given requested BER used as the QoS parameter and for each specific spreading factor SF, it is now straightforward to find the sub-carriers whose error rate is lower than the target BER (indicated by the circles). As soon as the sub-carrier BERis higher than the target, we just have to insert a new pilot symbol on this sub-carrier in order to refresh the information concerning the channel.

Thus, when a session with a terminal is initiated, a predetermined maximum allowable BER for the session is known. A set of subcarriers, a spreading sequence, a code rate and a modulation rate are initially assigned 2() for the session by the base station. Using the uplink SNR readings sent from the terminal to the base station, the base station calculates (from Equations I and 8) selects the pilot tone spacing that places the selected subcarriers close enough to cause the BER to lie below the maximum allowable BER, as illustrated in Fgure 6. Then it sends a control downlink signal to the terminal specifying the session parameters, including an indication of the pilot pattern selected. The session then proceeds with those parameters.

Periodically, the base station re-checks whether the BER is still acceptable using those session parameters, and if not, it reduces the density (spacing in time and/or frequency) of the pilot pattern, and signals to the terminal to commence using the new pilot pattem.

Adaptive Pilot Pattern As described above, for a given requested 'QoS', that means in our case, given a BER target, and for each spreading factor (SF=L), we can find the optimal conrclation factor p by means the theoretical relation of equation (8).

Once this value is obtained, it is then straightforward to get its correspondence in teems of frequency (and/or time) distance according to equation (2). This corresponds to updating rules for pilot tones, and finally to a flexible and adaptive pilot pattern.

In Figure 7 a qualitative example for an adaptive pilot pattern is indicated for a particular spreading factor. Figures 7A to D indicate the qualitative behaviour for four different (ascending) Quality of Service parameters, such as the bit error rate.

The progressive and regular appearance of pilot tones can be noted as the Hit Error Rate deteriorates. Also, In the case of severely degraded conditions of transmisson (see for example the eut-off in Figure 7A for a signal to noise ratio SNR<12 dB) no solution is possible, and thus no transmission is foreseen.

Second Embodiment: Adaptation Strategy based on Spreadine Factor and Pilot Pattern As described above with respect to equation (8), by varying the spreading factor, the transmission performance may be varied.

Figure 8 illustrates schematically the qualitative impact of the spreading factor on the link efficiency. The curves show the so-called overhead in dependency of the signal to noise ratio for different spreading factors. The overhead is defined as the number of pilot tone sub-carriers divided by the number of data tone sub-carriers.

It is clear that for a large overhead rate transmission becomes ineffective in terms of capacity.

A more efficient adaptation scheme according to this embodiment is therefore based on a joint adaptation of both spreading factor and pilot pattern. Thus, according to the required bit error rate and a certain permitted maximum overhead, the spreading factor and the pilot pattern are changed at the same time.

Figure <) illustrates schematically the qualitative behaviour of a joint adaptation scheme based on pilot tones and spreading factors.

The four diagrams of Figure 9A to D show the sub-carrier index as a function of the signal to noise ratio for four different target BERs. The dark and light lines indicate the behaviour for a low and a high spreading factor, respectively.

It can be seen from Fig. 9 that the spreading factor is increased when the maximal allowed overhead is reached.

Third Embodiment: Convolutional Codina of MC-SS Systems The above embodiments show how turbo-coding can be taken into account, but the same principles can be applied to other forms of error correcting encoding, as will be illustrated by the following embodiment.

According to John G. Proakis, Digital Communications, 3'0 edition, McGrawHill International Editions, chapter 8-2-3, (pp. 486-488), the following relation gives the Bit Error Probability for any convolutional code, depending both on its transfer function, and on the pairwise error probability: +0 Ph< I,BI! P2(d) (13) d =d /, whereby P2(d) is the pairwise error probability, and,B,, = a,' f(d), with a,'="numDer of paths of distance d from all ero path that merge with the all-zero path for the first time".

These parameters are directly related to the trellis structure of the code itself, and to the transfer function: +a' T(D, N) = ad ' Dd Nf (d) d -d /re.

Thus, equation (13) can be directly applied to convolutional coding, as our key function involving reception diversity (MRC) and correlation coefficient p is the pairvvise error probability Pa (d) . Then, the resulting BER is given by equation (13) for the MC-SS System using Convolutional Coding, and by equation (8) for Turbo-coding.

Using equation (13) instead of equation (8), a strategy for adapting pilot pattern or jointly adapting pilot pattern and spreading factors can be derived in the same manner as explicitly shown above for the case of the turbo codes.

Whilst in the above described embodiments the BER has been used as a parameter to describe the QoS, it is appreciated that alternatively other parameters suitable for describing QoS, such as for example SNR may be used.

Whilst in the above described embodiments Turbo-Codes and Convolutional codes have been described, it is appreciated that alternatively other coding schemes may be used.This invention may be used together with a method of uplinkmg channel quality data as described in our co-pending UK application GB04 (agent's reference J00046532GB, filed on the same day as the present application), and/or with a method of channel allocation as described in our co-pending UK application GB04 (agent's reference J00046534GB, filed on the same day as the present application), both of which are incorporated herein in their entirety.

Whilst the control of pilot patterns and spreading factors has been described herein, it will be clear that the analysis presented here (specifically, the derivation of a tightly calculated upper error bound by use of the estimate of the imperfection of the channels, for example due to spacing from pilot signals) can be used for other applications. It allows the analysis of the performance of any error correcting code (such as the Turbo Codes and Convolutional Codes presented herein) and thus serves as an efficient design algorithm to assess the robustness of such codes. All such other applications are within the scope of the invention.

It is to be understood that the embodiments described above are preferred embodiments only. Namely, various features may be omitted, modified or substituted by equivalents without departing from the scope of the present invention. The present invention extends to any and all such variants, and to any novel subject matter or combination thereof disclosed in the foregoing.

Claims (24)

1. Claims 1. A method of communication, wherein multiple data streams are

   transmitted simultaneously and channel coded a predetermined code, wherein the transmission performance is adapted by varying the pilot pattern; and wherein a relationship is used for estimating the transmission performance taking into account the accuracy of the channel model to adapt the pilot pattern.
2. 2. A method according to claim 1, wherein the time and/or frequency spacing of the pilot pattern is varied.
3. 3. A method according to claim 1 or 2, wherein multi-carrier spread spectrum transmission with channel coding is used.
4. 4. A method according to any preceding claim, wherein said predetermined code is turbo-coding.
5. 5. A method according to any of claims 1 to 3, wherein said predetermined code is convolutional coding.
6. 6. A method according to any preceding claim, wherein said relationship for estimating the transmission performance takes into account the imperfect channel estimates.
7. 7. A method according to any preceding claim, wherein the transmission performance is specified by a minimum bit error rate.

   8. A method according to any preceding claim, wherein said relationship for estimating the transmission performance takes into account maximal ratio combination.
8. 8. A method according to any preceding claim, wherein said relationship for estimating the transmission performance takes into account imperfect channel estimates.
9. 9. A method according to claim 8, wherein the imperfect channel estimate is introduced by a correlation factor describing the correlation between the pilot channel estimates and the true channel values.
10. 10. A method according to any preceding claim, wherein said relationship for estimating the transmission performance includes the average pair wise error probability.
11. A method according to any preceding claim, wherein frequency spreading is used.
12. 12. A method according to claim 10 or 11, wherein the transmission performance Is adapted by jointly varying the pilot pattern and the spreading factor of the frequency spreading.
13. 13. A method of communication, wherein data are used which are channel coded with a predetermined code and frequency spreading is applied, wherein the transmission performance is adapted by varying the pilot pattern and a spreading factor of the frequency spreading, and wherein a relationship is used for estimating the transmission performance taking into account the accuracy of the channel estimates at the l 5 reciever and the spreading factor to adapt the pilot pattern and the frequency spreading.
14. 14. A transmitter apparatus for transmitting data in a communications system, comprising means for channel coding the Incoming data with a predetermined code; means for multiplexing and spreading the coded data stream over a plurahty of subcarricrs; means for applying pilot tones; means for varying the transmission performance of the apparatus, wherein the transmission performance is adapted by varying the pilot pattern, based on the uplink signal to noise measurements from the reciever and on the distance of each subcarrier from a pilot tone.
15. 15. A transmitter apparatus for transmitting data in a communications system, comprising means for channel coding the incoming data with a predetermined code; means for multiplexing and spreading the coded data stream; means for applying pilot tones; means for varying the transmission performance of the apparatus, wherein the transmission performance is adapted by varying the pilot pattern and the spreading factor of the frequency spreading, and wherein a relationship is used for estimating the transmission performance under consideration of the accuracy of the channel model to adapt the pilot pattern and the spreading factor.
16. 16. A method of communication, including the step of estimating the perfonnance of turbo coded multi-carrier spread spectrum systems, taking into account maximal ratio combination and imperfect channel estimates.
17. 17. A method according to claim 16, wherein an expression for the pair wise error probability is used.
18. 18. A method according to claim 16 or 17, wherein the imperfect channel estimates are considered using a corTclation factor describing the correlation between the pilot channel estimates and the true channel values. s
19. 19. A method according to claim 16, 17 or 18, wherein the pair wise error probability is given by y7(i,.i2. ,iL.L,dp) J. (hi) (8)
20. 20. A method of controlling the session parameters of a communication session over a multicarrier communications link from a transmitter to a reciever, over which downlink data is carried in encoded form and spread over a selected plurality of subcarricrs which also carry pilot signals spaced in time and/or frequency, comprising selecting the number of said subcarriers and the spacing of said pilot signals using a relationship between quality of service, spacing of the selected subcarriers from the pilot signals, number of subcarTicrs and parameters of the encoding.
21. 21. A method of controlling a coded multica'Tier tclecoTllmunications system \hicl1 includes use of a measure of the quality of service (for example, BER) based on a model of the errors on individual subcarriers.
22. 22. A method of controlling a multicarrier telecommunications system which includes use of a measure of the pairwise error probability of a code based on a model of the errors on individual subcarriers.
23. 23. A method according to claim 21 or claim 22, in which the model calculates the subcarrier imperfections based on their spacing from a pilot signal.
24. 24. A transmitter system for use in a communications system, adapted to implement the method set out in any of claims I to 13 or 16 to 23.