GB2508942A - Reducing the peak-to-average ration in an OFDM system using a unitary matrix transformation - Google Patents

Abstract

To reduce the peak-to-average power ratio (PAPR) in an orthogonal frequency division multiplexed (OFDM) based communications system, a transmitter applies a unitary matrix transformation 340 to modulated symbols corresponding to input data streams 320 before the transformation of the symbols by an inverse Fast Fourier Transformation (IFFT) algorithm 308. Pilot symbols are inserted using a multiplexer 342 prior to the IFFT processing. In another embodiment (figure 4A) the pilot symbols also undergo a unitary matrix transformation prior to insertion. In further embodiments supplementary post-IFFT PAPR reduction techniques are applied, for example, active constellation extension (ACE) or tone reservation (TR) algorithms (see figures 6A and 7A). Corresponding receiver arrangements are also disclosed.

Description

DATA PROCESSING APPARATUS AND METflOI) The present disclosure relates to transmitters and methods for transmitting data in a wireless communications system. In particular, the disclosure relates to the reduction of Peak-to-Average Power Ratio (PAPR) in multi-carrier transmission systems. The disclosure also relates to receivers and methods for detecting and recovering data from transmissions generated by such transmitters and transmission methods.

Multi-carrier transmission systems (and, in particular, systems based on Orthogonal Frequency Division Multiplexing, OFDM) offer many advantages over single carrier systems. Notably they offer greater robustness to multi-path fading and inter-symbol interference. These strengths lie behind the adoption of multi-carrier techniques in many world-wide broadcast and telecommunications standards such as Digital Video Broadcasting (DVB), ISDB-T, DTMB, WiMAX, LTE etc. Systems which have been arranged to operate in accordance with DVB standards (which span terrestrial and cable broadcast) for example, utilise OFDM. OFDM can be generally described as providing IV narrow band sub-carriers (where N is an integer) winch are modulated in parallel, each sub-earner communicating a modulated data. symbol such as Quadrature Amplitude Modulated (QAM) symbol or Quadrature Phase-shift Keying (QPSK) symbol. The modulation of the sub-carriers is formed in the frequency domain and transformed by an inverse fast Fourier transform (lEFT) algorithm into the time domain for transmission. Since the data symbols are communicated in parallel on the sub-carriers, the same modulated symbols may be communicated on each sub-carrier for an extended period. The sub-carriers are modulated in parallel contemporaneously, so that in combination the modulated carriers form an OFDM symbol. The OFDM symbol therefore comprises a pluralit' of sub-carriers each of which has been modulated contemporaneously with different modulation symbols.

One drawback of OFDM, with respect to single-carrier technologies, is the comparatively high Peak-to-Average Power Ratios (PAPR). PAPR is defined as the ratio of the maximum power to the average power of a signal during an OFDM symbol period.

PAPR becomes an linportant factor when building the transmission infrastructures to support OFDM-based systems or when trying to switch the present infrastructure fi-om supporting a single-carrier to a inulti-canier teelmo1o'. The PAPR values increase as the number of carriers increases: as the trend is towards deploying transmission systems with higher carrier numbers due to spectral efficiency gains, the situation is predicted to become worse.

Since most practical transmission systems are peak-power limited, one way of alleviating this problem is lo reduce the average transmit power to levels well below the peak power available (this might he informally expressed as having "large back-offs" in the system) to ensure linear operation over the ifill dynamic range. This, in turn, could result in reduced received signal power which is not a desired outcome.

In practice, to avoid large back-offs, occasional saturation of the power amplifiers or clipping in the digital-to-analogue converters may be tolerated. This non-linear process creates inter-modulation distortion that could lead to increased hit error rate (HER) in a standard linear receiver. It also causes spectral widening of the transmit. signal: leading to increased susceptibility to adjacent channel interference.

Several methods have been devised either to reduce the PAPR levels or to combat its effects in the transmitter and/or receiver.

In a paper entitled "Reducing the peak-to-average power ratio using unitary matrix transformation" by Zhu, et al. [JET Commun.. 2009, Vol. 3, Iss. 2, pp 161-171] the authors discuss a theoretical scheme for reducing the PAPR in an OFDM based communication system. They describe the application of a Unitary Matrix Transformation (IJMT) prior to the building of OFDM symbols with the S resulting reduction iii PA PR.

Practical implementations of OFDM transmission systems rely upon pilot signals (frequency-domain pilot carriers) to enable robust channel estimation and to facilitate low-complexity channel equalisation at the receiver. Pilot signals however are necessarily repetitive and periodic in nature giving rise to an increased likelihood of a higher PAPR.

Summary of Disclosure

According to one aspect of the disclosure, there is provided a transmitter for transmitting data in a wireless communications system, the transmitter comprising: a converter having circuitry configured to convert input modulated symbols into respective data vectors; a unitary matrix transformer unit having circuitry configured to transfonn the data vectors into colTesponding transformcd vcctors; a multiplexer having circuitry configured to receive input modulated pilot symbols and the transformed vectors and to output a plurality of multiplexed transformed vectors; and an Orthogonal Frequency Division Multiplexing, OFDM, symbol generator having circuitry configured to generate an OFDM symbol from the multiplexed transformed vectors, the OFDM symbol including a plurality of sub-carrier symbols formed in the frequcncy domain and transformed by inverse fast Fourier transform, IFFT, into a plurality of samples in the time domain for transmission.

According to a further aspect of the disclosure, there is provided a method for transmitting data in a wireless communications system, the method comprising: converting input modulated symbols into respective data vectors; transforming by circuitry the data vectors into corresponding transformed vector according to a unitary matrix transfonnation; receiving input modulated pilot symbols; multiplexing the transformed vector with infonnation corresponding to the received modulated pilot symbols to output a plurality of multiplexed transformed vectors; and generating by circuitry an OFDM symbol from the multiplexed transformed vectors, the OIDM symbol including a plurality of sub-carrier symbols formed in the frequency domain and transformed by inverse fast Fourier transfonn, IFFT, into a plurality of samples in the time domain for transmission.

Various aspects and features of the present disclosure are defined in the appended claims.

Further aspects of the present invention include a method of detecting and recovering data from Orthogonal Frequency Division Multiplexed (OFDM) symbols.

Brief Description of Drawings

Embodiments will now be described by way of example only with reference to the accompanying drawings, wherein like parts are provided with corresponding reference numerals, and in which: Figures 1A and lB are schematic block diagrams of respective simplified OFDM transmitter and

receiver paths in accordance with the prior art;

Figures 2A and 2B are schematic block diagrams of respective simplified OFDM transmitter and receiver paths having a UMT prior to OFDM symbol generation in the transmitter path in accordance

with the prior art;

Figures 3A and 3B are schematic block diagrams of respective simplified pilot-aided OFDM transmitter and rcccivcr paths having a IJMT prior to OFDM symbol generation in the transmitter path in accordance with one embodiment Figures 4A and 4B are schematic block diagrams of an alternative pair of respective simplified pilot-aided OFDM transmitter and receiver paths having a 1JMT prior to OFDM symbol generation in tile transmitter path in accordance with a further embodiment; Figures 5A and 5B are schematic block diagrams of respective simplified pilot-aided OFDM transmitter and receiver paths having a UMT prior to OFDM symbol generation and a post-lEFT processing PAI'R reduction step in the transmitter path in accordance with another embodiment; Figures 6A and ÔB are schematic block diagrams of respective simplified pilot-aided OEDM transmitter and receiver paths having a UMT prior to OFDM symbol generation and a post-IFFT TR step in the transmitter path in accordance with yet another embodiment; Figures 7A and 713 are schematic block diagrams of respective simplified pilot-aided OFDM transmitter and receiver paths having a UMT prior to OFDM symbol generation and a post-lEFT ACE step in the transmitter pad in accordance with a further embod iinent; Figures SA and 813 show the respective signal magnitudes of Original and IDFTI'ransforrned Signals; Figure 9 compares the PAPR Performance of Original and IDFT-fransformed Signals; Figure 10 conftasts the in-baud characteristics of the system of Figure 2 with those of Figure 1 by comparing BER Performance of Original and IDFT-Transformcd Signals over an additive white Gaussian noise (AWGN) Channel; Figure 11 contrasts the in-hand characteristics of the system of Figure 2 with those of Figure 1 in channels heavily distorted by fade (i.e. strong multi-path conditions) by comparing the BER Performance of Original and IJ)FT-Transformed Signals over a Frequency-Selective Channel with 3% Erasures; Figure 12 contrasts the out-of-band characteristics of the system of Figure 2 with those of Figure 1 by comparing die power spectral density (P50) of Original and IDFT-Transformed Signals; Figure 13 compares the Absolute Aperiodic ACF Measurements of: i) Original OFDM; ii) IDFT-Transformed OFDM with pilots carriers (as in Figures 3A13B); and iii) IDFT-Transformed OFDM without pilots carriers (as in Figure 2); Figure 14 compares PAPR Perfonnance of Original and IDFT-Transformed OFDM -with and without pilot carriers; Figure 15 contrasts the out-of-hand characteristics of the systems of Figures 1, 2 and 3A/3[3 by comparing PSD of Original and lUFF-Transformed OFDM -with and without pilot carriers; Figure 16 compares the Absolute Aperiodic ACE Measurements of Original and IDFT-Transformed OFDM -with and without pilot carriers: two different pilot schemes are shown, representing a high-density pilot population (PP1) & a low-density pilot population (PP7) respectively; Figure 1 7 compares PAPR Perfonnance of Original and IDFT-Transfortned OFDM -with and without pilots carriers: with two different pilot schemes PP1 and PP7; Figure 18 compares the Absolute Aperiodic ACE Measurements of Original and IDFT-Transformed OFDM in accordance with the embodiment of Figure 4A; Figure 19 compares the Absolute Aperiodic ACE Measurements of IDFT-Transformcd OFDM in thc schemes of Figure 3A and Figure 4A; Figure 20 compares the PAPR Performance of Original OFDM and IDFT-Transformed OFDM in accordance with the embodiment of Figure 4A; Figure 21 compares the PAPR Performance of IDFT-Transformed OFDM in the schemes of Figure IA and Figure 4A; Figure 22 compares the Absolute Aperiodic ACF Measurements of Original OFDM and the IDFT-Transformed OFDM scheme of Figure 6A (combining the UMT technique and the TR post-IFFT PAPR reduction technique); Figure 23 compares the PAPR Performance of Original OFUM and the IDFT-Transformed OFDM scheme of Figure 6A (with and without UMT and/or TR); Figure 24 compares the Absolute Aperiodic ACE Measurements of Original OFDM and the IDFT-Transformed OFDM scheme of Figure 7A (combining the UNIT technique and the ACE post-IFFT PAPR reduction technique); Figure 25 compares the PAPR Perfonnance of Original OFDM and the IDFT-Transformed OFDM scheme of Figure 7A (with and without UMI and/or ACE); Figure 26 illustrates successive operations in a OFDM transmitter path having UMT; Figure 27 illustrates the PAPR Performance of Original and IDFT-Transformed Signals with scatter pilots alone and with both scatter pilots and continual pilots; Figures 28A and 28B allow the comparison of the Absolute Aperiodic ACF Measurements for pilot-less simulations (Figure 28A) and pilot-aided simulations with different combinations of SP and CP (Figurc 28B). in each case with and without the application of an IDFT-Transfonn (UMT); Figure 29 shows a schematic block diagram of a transmitter in an enhanced Pilot-Assisted UMT-OFDM System; Figure 30 shows a schematic block diagram of a receiver for an enhanced Pilot-Assisted UMT-OFDM System; Figure 3 1 shows the Complimentary Cumulative Distribution Function (CCDF) indicative of the PAPR perfonnance for both simplified and enhanced sftuelnrcs when boosted SF and CF carriers are inserted; Figure 32 shows a schematic block diagram of a transmitter in an enhanced Pilot-Assisted UMT-OFDM System modified to accommodate a post-WET scheme; Figure 33 shows the PAPR performance of the combined pre-and post-WET processes with Tone Reservation (TR) for the enhanced system of Figure 32; and Figure 34 shows the PAPR performance of the combined pre-and post-WET processes with Acl.ive Constellation Extension (ACE) for the enhanced system of Figure 32.

Description of Embodiments

Figures 1A and lB illustrate a simplified model of the transmit and receive paths in an OFDM scheme respectively.

In the transmit path (Figure 1 A), input bit streams 120 are mapped to modulated data symbols (in this case, for example, Quadrature Amplitude Modulated (QAM) symbols; an alternative is Quadrature Phase-shift Keying (QPSK) symbols) by a mapping unit 102. For the purposes of the following description, units indicated as "QAM mapping units" and "QAM demapping units" may be considered to represent any suitable mapping/demapping unit including units that use a QAM or QPSK modulation scheme. These modulated symbols are applied in parallel (having been converted from serial to parallel by a. S/P converter 104) to an OFDM generation module 110.

The OFDM generation module 110 includes an edge carrier insertion unit 106, an Inverse Fast Fourier Transfer (WET) arithmetic operation unit 108, and a guard interval (GD insertion unit 112. Tn the OFDM generation module 110 of Figure IA, the modulated symbols are allocated to plural sub-carriers, respectively, each of the symbols being transformed by the hiverse Fast Fourier Transfer (IFFT) arithmetic operation 1 08.

In the corresponding receive path (Figure IB), the demodulation of the OFD.M signal obtained through the modulation can be carried out by using a Fast Fourier Transfer (FF1) arithmetic operation 154 for carrying out Fourier transfonn.

The receive path includes a guard interval removal unit 152, Fast Fourier Transfer (FFT) arithmetic operation unit 154, a channel estimator unit 156, a channel equalizing unit 158, an edge carrier removal unit 1 60, a parallel to serial converter 162 and a demapping unit 164. Received signals are thus subjected to a process which is in many respects the inverse of the transmit path. A transmitting apparatus which transmits the OFDM signal can be configured by using a circuit for carrying out the IFFT arithmetic operation. Also, a receiving apparatus which receives the OFDM signal can be configured by using a circuit for carrying out the FF1 arithmetic operation.

In addition, in the OFDM system, provision of signal intervals called guard intervals by the guard interval (GT) insertion unit 112 makes it possible to enhance the resistance against multi-path interference.

A eat variety of devices may be provided with the necessary hardware to implement a receiver path such as that in Figure lB. Instances of such receiving devices include: TV panel, set top boxes (STB), recorder device (personal video recorders. PVRs), home gateways, streamers, mobile handsets, general purpose computing devices in a range of form factors (including desktop, laptop, netbook, tablet and PDA computers). In many devices, it is convenient to provide a plurality of receivers (and tuners) to facilitate plural reccive activities substantially in parallel. Furthermore, the hardware for implementing a receiver path may he provided in a peripheral device which cooperates with a general purpose computin.g device; for example an auxiliary DVB-T or ATSC receiver for connecting via a wired or wireless interface (such as USB. 1EEE1394 or other proprietary interconnection or docks).

An ideal' PAPR reduction process has favourable transmission/reception characteristics including, for example: * Reasonably high' PAPR reduction values (perhaps exceeding 3 dB compared to signal with no PAPR reduction process applied), * No loss of data capacity as a result, * No signal power increases, * No degradation of out-of-band (OOB) characteristics, * No degradation of in-band characteristics, * No channel noise (interference) magnifications, * Moderate overhead on transmitter and receiver complexities, S * Works well in cascadcd transmission chains, * No incompatibility with any of the possible transmission or receiver configurations.

A good practical solution is thus one that achieves at least some of these targets.

As noted above, one known scheme for reducing IPAPR in OFDM transmission is described in a paper entitled "Reducing the peak-to-average power ratio using unitary matrix transformation" by Zhu, et al. Figures 2A and 28 shows a simulation model of the Zhu scheme. The scheme uses lIME to reduce the correlation properties of the signal entering the IFFT process. This in turn leads to reduced PAPR at the output of the OFDM generating block.

[he paper observes that there is a close relation between the PAPR of an OFDM signal and the aperiodic mitocorrelation function (ACE) of the input vector -in a simple model, the input vector is 1 5 generated by a mapping unit (such as the QAM Mapper 202 in Figure 2A) and the components, X1. are complex.

For an input vector, X, define the aperiodic autocorrelation function (ACE) as: PQ) = L0 Xk÷IX;,/ = 0,1...., N -l (where superscript * denotes complex conjugate and Nis the number of subcarriers in the OFDM system) 1 he absolute aperiodic AF function is It can be shown that the PAFR of input vector Xis directly proportional to A, where A is defined

N-I

as: A. = The paper's authors reason that reducing the absointe aperiodic ACE (i.e. making A. as small as possible) will reduce the PAPR levels. They suggest that applying a suitable transfonnation to data entering the lEFT process (i.e. one that lowers the absolute aperiodic ACE of this data) guarantees a reduced PAPR at the output of the WFT. One class of transformations, Unitmy Matrix Transformations (UMT) is identified as suitable by the authors. They then explain their scheme using one particular UMT -the unitary form of the inverse discrete Fourier Transform (IDFT). A unitary matrix transformation is a transformation that preserves the scalar product.

This approach is demonstrated to cause no degradation of out-of-band (OOB) characteristics (tested in Figure 2A by comparing the power spectral density (PSD) of a signal with tJMT applied before lEFT and that for a signal without UMT). Furthermore, the approach can be seen to cause no degradation of in-band characteristics either (tested in Figure 2A by comparing bit error rate (BER) performance for a signal where no forward error correction (FEC) coding is applied with and without the application of UMT prior to IFFT).

In the transmit path (Figure 2A), input bit streams 220 are mapped to modulated data symbols (for consistency with Figure IA, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 202. These modulated symbols are applied in parallel (having been converted from serial to parallel by a.

S/P converter 204) to a unitary matrix transformer unit 240 prior to transmission to an OFDM generation module 210.

As for Figure 1A, the OFDM generation module 210 includes an edge catTier insertion unit 206, an Inverse Fast Fourier transfer (lEFT) arithmetic operation unit 208, and a guard interval (GI) insertion unit 212. Tn the OFDM generation module 210 of Figure 2A, the modulated symbols are allocated to plural sub-carriers, respectively, cacti of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 208.

In the corresponding receive path (Figure 2B), the demodulation of the OFDM signal obtained through the modulation can he carried out by using a Fast Fourier Transfer (FFT) arithmetic operation 1 0 254 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 252, Fast Fourier Transfer (FFT) arithmetic operation unit 254, a channel estimator unit 256, a channel equalizing unit 258, an edge carrier removal unit 260, an inverse unitary matrix transformer unit 270, a parallel to serial converter 262 and a demapping unit 264.

Figure 26 illustrates successive operations in an OFDM system transmitter path with UMT. Input data bits 2602 are first modulated by a mapping unit 2612 (exemplary rnodulatioiis include MQAM and QPSK). Blocks of the resulting modulated symbols 2604 are converted from serial to parallel by a. S/P converter 2614 parallel symbols, now assembled as coefficients of data vectors 2606, being represented as a plurality of arrows. A unitary matrix (such as a unitary form of the inverse discrete Fourier transform, IDFT, matrix) is multiplied by the data vectors 2606 in a UMT unit 2616. The transformed vectors 2608 generated by the TJMT unit 2616 are then applied to an IFFT Process 2618 to generate an OFDM symbol 2610. A transmission signal may be assembled from a plurality of such OFDM symbols.

Zhu et al. proposed an algorithm that showed promising results in a pilot-less OFDM transmission system. The Zhu paper does not however consider the more practical problems arising from the presence of frequency-domain pilot carriers (pilot signals) -a feature of many (JFDM-based communication systems (including nearly all current and emerging wireless standards).

Frequency-domain pilot carriers are typically provided to enable robust channel estimation coupled with low-complexity channel equalization at the receiver. They arc also used for measuring and compensating for timing errors, frequency offsets and common-phase errors, for example.

Conventionally, two types of pilots are provided: scattered pilots and continual pilots. Scattered pilots are uniformly spaced among the carriers in any given symbol. In contrast, the continual pilot signals occupy the same carrier consistently from symbol to symbol. .Pilot signals of both types are typically generated by a pilot generation unit.

The present inventor has recognised the impact of frequency-domain pilot carriers upon the UMT based scheme described above and extended this scheme to pilot-aided OFDM transmission systems.

Three implementations of this extended scheme are considered.

Figures 3A and 3B show block diagrams of simplified pilot-assisted OFDM transmission and receiver systems respectively, in these and subsequent systems, the presence of pilot carm-iers is modelled using only scattered pilots (where the context does not require both continual pilots and scattered pilots).

The presence of both scattered pilots and continual pilots (typical of the majority of conventional OFDM transmission systems) lowers any PAPR reduction gains from the application of UMT prior to lEFT to effectively zero.

It should be appreciated that the pilot carriers need to be accessed at the receiver input without any dependency on the data pay] oad. In fact, it is necessary to remove all channel artifacts before proceeding to any other receive path processes including inverse UMT transformation. This, therefore, requires that the pilot carriers bc excluded from any grouped UM'!' transformations and that they can only be inserted alter the data carriers undergo the UIMI transformation in the transmission path.

In the transmit path (Figure 3A), input bit streams 320 are mapped to modulated data symbols (for consistency with Figure 1A, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 302. These modulated symbols are applied in parallel (having been converted from serial to parailel by a S/P converter 304) to a unitary matrix transformer unit 340 prior to transmission to an OFDM generation module 310.

As for Figures IA and 2A, the OFDM generation module 310 includes an edge carrier insertion unit 306, an inverse Fast Fourier Transfer (li/Fl) arithmetic operation unit 308, and a guard interval (GI) insertion unit 312. In the OFDM generation module 310 of Figure 3A, the modulated symbols arc allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 308.

In this example, rahdomized boosted BPSK pilot data (output by a BPSK mapper 332) is inserted by a multiplexer 342 in-between (scattered) the UIMT transformed data at predetermined periodic intervals. It should be noted that this process can only take place after IJMT transfonnation 340.

Otherwise, the pilot-information would become payload-data dependent making the process of pilot-demodulation and consequently channel estimation impossible at the receiver.

In the corresponding rcccivc path (Figure 313), the demodulation of the OFDM signal obtained through the modulation can he carried out by using a Fast Fourier Transfer (FF1) arithmetic operation 354 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 352, Fast Fourier Transfer FFT) arithmetic operation unit 354, a channel estimator unit 356, a channel equalizing unit 358, an edge carrier removal unit 360, a demultiplexer 366, an invcrsc unitary matrix transformer unit 370, a parallel to serial converter 362 and a demapping unit 364.

It is clear that, pilots needed to be extracted to establish channel estimation/equalization, to remove channel distortions, before inverse U]\4T transformation 370 could be performed. Boosting of BI'SK pilot data is optional.

As explained previously, the UMT transfonnation 340 minimizes the correlations that exist in the QAM data leading to reduced probability of tones adding together to create large peaks during IFFT process 308. This means that maximum PAPR gains are achieved if UMI-transformed data entered the IFFT process 308 without any other changes/modifications. Otherwise, the autoeorrelation properties are interfered with and consequently we should expect reduced PAPR reductions.

In the embodiment of Figure 3k there are two additional processes in between the LTMT-transformation 340 and the IFFT process 308, namely pilot-insertion 342 and edge carrier-insertion 306.

Without edge-carrier-insertion, the OFDM frequency mask would be flat across the band, the edge carriers are thus inserted to allow frequency interpolation up to the edge of the OFDM spectrum to achieve the conventional high-roll off of the spectrum mask (at extreme edges). It should be noted here that reducing the number of edge carriers would result in improved PAPR performance hence extended-bandwidth mode with high FFT sizes is a preferred mode of operation.

Pilot signals (signals of known values and frame positions within the OFDM symbol, known in advance by the receiving apparatus side) arc discretely inserted into a time direction an&or a frequency direction. I'he receiving apparatus (Figure 3B) utilizes the pilot signals in synchronization, and estimation of the transmission path characteristics 356.

Periodic pilot insertion inherently leads to some degree of degradation of the autocorrelation properties of the UMT signal resulting in increased absolute aperiodic ACF values and, following the argument set above, reduced PAPR performance, with rcspcct to pilot-less OFDM can be expected, when pilot-carriers are inserted. The pilot carriers are usually transmitted with boosted signal powers, to provide them with additional robustness against channel impairments. This increased, relative power leads to a further drop in PAPR reduction gains.

The present inventor has however leant that the degradation is not as profound as might have been predicted.

In a further embodiment, a slight modification of the embodiment of Figures 3A and 313, modulated (QAM) data corresponding to payload 420 and pilot signals 430 are passed through separate UMI-transformation units 440, 436 and then mixed 442 before entering the IFFT process 408. This embodiment is illustrated in Figures 4A and 4B. Comparison of the performances of the embodiments of Figures 3A13B and Figures 4A14B revealed a small improvement in the latter over the former.

Figures 4A and 4B show block diagrams of the transmit and receive operations, respectively, in an embodiment where QAM data 402 corresponding to payload 420 and pilot signals 430 are passed through separate UMT-tt-ansfonnation units 440, 436.

In the transmit path (Figure 4A), input bit streams 420 are mapped to modulated data symbols (for consistency with Figure IA, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 402. These modulated symbols are applied in parallel (having been converted from serial to parallel by a S/P converter 404) l.o a unitary matrix transformer unit 440 prior to transmission to an OFDM generation module 410.

As for previously described transmit paths, the OFDM generation module 41 0 includes an edge carrier insertion unit 406, an Inverse Fast Fourier Transfer (IFFT) arithmetic operation unit 408, and a guard interval (GI) insertion unit 412. In the OFDM generation module 410 of Figure 4A. the modulated symbols are allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 408.

In this example, randomized boosted BPSK pilot data (output by a BPSK mapper 432) is inserted by a multiplexer 442 in-between (scattered) the UMT transformed data at predetermined periodic intervals. The pilot data here is parallelised by a S/P converter 434 and applied to a separate UMT-transformation units 436, prior to insertion by the multiplexer 442.

hi the corresponding receive path (Figure 4B), the demodulation of the OFDM sigral obtained through the modulation can be carried out by using a Fast Fourier Transfer (FFT) arithmetic operation I: 454 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 452, Fast Fourier Transfer (FFT) arithmetic operation unit 454, a pilot inverse UMT-transformation unit 472, a channel estimator unit 456, a channel equalizing unit 458, an edge carrier removal unit 460, a demultiplexer 466, an inverse unitary matrix transformer unit 470, a parallel to serial converter 462 and a demapping unit 464.

IJMT transfonnation 436 of the pilot signals may be performed by a dedicated pilot signal unitary matrix transformer unit or alternatively by the same unitary matrix transformer unit used to perform UMU-transformation 440 on the payload signals.

This embodiment also requires additional processing in the receiver path: the extracted Uh'll'-transformed pilots have to undergo a separate inverse-UMT transformation 472 before being used in the channel estimation process 456. Again, this separate inverse-UMT transformation 472 may be carried out by a dedicated invcrsc-UMT transformer unit or by the same inverse-UMT' transformer unit used to perform inverse-IJN4T transformation 470 on the received payload data.

Generally, the small improvement in measured PAPR reduction may not justi' the additional complexity both at the transmitter and receiver in a practica.l system.

In combination with the embodiments of Figures 3A and/or 4A, further embodiments are now described where a supplemental post-IFFT PAPR reduction algorithm 515 is applied. For the purposes of illustration, these further embodiments are described as extensions of the scheme illustrated in Figure 3A: nonetheless the reader will readily appreciate that it is possible to extend the embodiment of Figure 4A in a similar manner.

IJMT-transforrned PAPR reduction is a prc-IFFT method with zero-loss of complexity and reasonably small order of complexity. The present inventor recognised that it is possible to increase still furlher the PAPR reductions by including a post-WV!' PAM reduction algorithm 515 to compensate partially for the loss of PAPR performance due to pilot-insertions 542.

Figures 5A and SB show the application of a supplemental post-IFFT PAPR reduction algorithm to the scheme illustrated in Figure 3A13B.

In the transmit path (Figure 5A), input bit streams 520 are mapped to modulated data symbols (for consistency with Figure lA, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 502. These modulated symbols are applied in parallel (having been converted from serial to parallel by a S/P converter 504) to a unitary matrix transformer unit 540 prior to transmission to an OFDM generation module 510.

As for previously described transmit paths, the OFDM generation module 510 includes an edge carrier insertion unit 506, an Inverse Fast Fourier Transfer (IFFT) arithmetic operation unit 508, and a guard interval (CI) insertion unit 512. In the OFDM generation module 510 of Figure 5A, the modulated symbols are allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 508.

In this example, randomized boosted BPSK pilot data (output by a T3PSK mapper 532) is inserted by a multiplexer 542 in-between (scattered) the UIMT transformed data at predetermined periodic intervals.

A post-Wl"F PAPR reduction scheme is then applied by a unit 515.

In the corresponding receive path (Figure SB), the demodulation of the OFDM signal obtained through thc modulation can he carried out by using a Fast Fourier Transfer (FFT) arithmetic operation 554 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 552, Fast Fourier Transfer (FFT) arithmetic operation unit 554, a channel estimator umt 556, a channel equalizing unit 558, an edge carrier removal unit 560, a demultiplexer 566, an inverse unitary matrix transformer unit 570, a parallel to serial converter 562 and a demapping unit 564.

In principal, in a pilot-aided OFDM system, the UMT-based transformation provides up to 2.0 dB PAPR improvements as long as the correlation properties of UMT-transfonned data, before entering the ffFT, are not interfered with in such a way that AA-ACF values were magnified. Thus, any post-JFFT PAPR reduction technique 515 that respects this condition can be deployed to increase the PAPR performance.

Two suitable exemplary candidates for post-lEFT PAPR reduction algorithms are known from the DVI3-'1'2 ETST Specification: Tone Reservation(TR) and Active Constellation Extension (ACE).

1 0 Embodiments using TR and ACE (respectively) in addition to the UMT (pre-WFT process) arc set out below.

The reader will readily appreciate that these known algorithms are merely illustrative of the class of post-[FFT processes which might be applied to the arrangements of the earlier-described embodiments.

TR and ACE may be applied separately and, in many cases, in combination. Other post-WET PAPR reduction techniques, which respect the condition upon AA-ACF values set out above, may be adopted in place of, or in addition to, the ACE and TR techniques. The following embodiments are described in terms of these known algorithms however it is not intended that these embodiments represent a limitation to the use of these precise algorithms (either individually or in combination) only.

Considering the case where TR is used as the post-IFFT process, the PAPR reductions are mostly due to those offered by the TR only. II has been determined that the PAPR gains are not cumulative. It appears that this is because the condition upon AA-ACF values is broken. Figure 6A illustrates aUMT system where the TR algorithm 615 has been applied.

In the transmit path (Figure 6A). input bit streams 620 are mapped to modulated data symbols (for consistency with Figure IA, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 602. These modulated symbols are applied in parallel (having been converted from serial to parallel by a S/P converter 604) to a unitary matrix transformer unit 640 prior to transmission to an OFDM generation module.

As for previously described transmit paths, the OFDM generation module includes an edge carrier insertion unit 606, an Inverse Fast Fourier Transfer (lEFT) arithmetic operation unit 608, and a guard interval (GI) insertion unit 612. In the OFDM generation module of Figure 6A, the modulated symbols are allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 608.

In this example, randomized boosted BPSK pilot data (output by a BF'SK mapper 632) is inserted by a multiplexer 642 in-between (scattered) the UMT transformed data at predetermined periodic intervals. The multiplexer also inserts dummy data (i.e. a reserved carrier incorporating a peak-reduction siwial) for use intone reservation -this dummy data being generated by a generator 644.

A post-lEFT PAPR reduction scheme is applied to the output of the ffFT arithmetic operation unit 608 by a tone reservation unit 615.

In the corresponding receive path (Figure 6B), the demodulation of the OM sigual obtained through the modulation can he carried out by using a Fast Fourier Transfer (FFT) aritlunetic operation 654 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 652, Fast Fourier Transfer (FFT) arithmetic operation unit 654, a channel estimator unit 656, a channel equalizing unit 658, an edge carrier removal unit 660, a demultiplexer 666, an inverse unitary matrix transformer unit 670, a parallel to serial converter 662 and a demapping unit 664.

An embodiment in which the ACE algorithm 715 is applied is illustrated in Figure 7A. Here the PAPR reduction gains are cumulative, resulting in a significant net gain over the application of UMT alone.

In the transmit path (Figure 7A), input bit streams 720 are mapped to modulated data symbols (for consistency with Figure IA, Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 702. These modulated symbols are applied in parallel (having been converted from serial lo parallel by a S/P converter 704) to a unitary matrix transfonuer unit 740 prior to transmission to an OFDM generation module.

As for previously described transmit paths, the OFDM generation module includes an edge carrier insertion unit 706, an Inverse Fast Fourier Transfer (IFFT) arithmetic operation unit 708, and a guard interval (Ci) insertion unit 712. In the OFDM generation module of Figure 7A, the modulated symbols are allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 708.

In this example, randomized boosted BPSK pilot data (output by a BPSK mapper 732) is inserted by a multiplexer 742 in-between (scattered) the UMT transformed data at predetermined periodic intervals.

A post-IFFT PANt reduction scheme is applied to the output of the IFFT arithmetic operation unit 70% by an Active Constellation Extension (ACE) unit 715.

In the corresponding receive path (Figure 7B). the demodulation of the OFDM signal obtained through the modulation can be carried out by using a Fast Fourier Transfer (FFT) arithmetic operation 754 for carrying out Fourier transform.

The receive path includes a guard interval removal unit 752, Fast Fourier Transfer ftFT) arithmetic operation unit 754, a channel estimator unit 756, a channel equalizing unit 758, an edge carrier removal unit 760, a demultiplexer 766, an inverse unitary matrix transformer unit 770, a parallel to serial converter 762 and a demapping unit 764.

The performance of the various embodiments discussed above are now compared to the performance of the prior art arrangement illustrated in Figures 2A12B.

To establish a baseline for comparison, the performance evaluation results as reported in Zhu, et aL are simulated. The block diagram shown in Figures 2A12B was modelled in MATLAB. The following simulation parameters were used (matching very closely values used in the paper): * FFT Size: 256 Number of active carriers: 213 * QAM: QPSK * Guard Interval: /4 FEC: None * Unitary Matrix Transformation: Inverse DFT * Oversampling Ratio: 4 * Channel Conditions:

-AWGN

-0dB Echo [3% erasure] * Channel Estimation: Ideal Channel Equalization: Zero-Forcing * Dc-mapping: Hard-Decision Figures SA and 8B show the signal amplitude (magnitude) of the original and the IDFT-transformed signals. The average signal magnitude has remained the same whilst the peak magnitudes are suppressed.

Figure 9 shows the PAPR performance of the original and the IDFT-transfonned signals through Complimentary Cumulative Distribution Function (CCDF) curves, In this set up, the UMT offers 4 dB improvements at the CCDF value of lxi o.

As set out above, the technique seeks to improve the PAPR performance whilst avoiding the disruption or diminishing of other characteristics of the newly formed signal. Some of these characteristics are considered here: the in-band characteristics are represented by measurements for un-coded BER performance and out-of-band characteristics represented by looking at the power spectral density (PSD) of the signal. The choice of un-coded OFDM (no FEC protection) would reveal any slight degradation as the results are not masked by the FEC encoder-decoder performances.

Figure 10 shows the BER performance of the original and the IDFT-transformed signals over the AWUN channel. [here is no degradation of the BER performance due to IJMT transformation.

Figure 11 shows the BER performance of the original and the IDFT-transfomied signals over a severely faded frequency-selective channel (0 dB echo channel') with a 3% erasure rate. In this case, the BER performance is, in fact, improved significantly as the transformation acts also as a cell interleaver'.

Figure 12 shows the PSD of the original and the IDFT-transformed signals. It clearly shows that the out-of-band characteristics remain unaffected.

The performance of a system when pilot-carriers are insefted is discussed below. Figures 3A and 3B depict a simplified structure for a pilot-assisted UMT-OFDM transmit-receive system.

Performance evaluation is extended to high FFT sizes. Performance was measured for both 32K and 64K modes, extended and non-extended bandwidth modes, pilot patterns PP1 and PP7 (as defined in DVB-T2 specification). In relation to the results of simulations in Figures 13 to 25, only the 32K-Extended-PPI results are discussed. The corresponding DVB-T2 parameters are utilized: * FFT Size: 32K [Extended-Bandwidth Mode] * QAM: QPSK * Guard Interval: 1/4 * FEC: None * Pilot Pattern: PP1[Dx3, Dy44DxtDy'12] * Unitary Matrix Transformation: Inverse DFT * Oversampling Ratio: 4 Figures 13 -15 show the absolute aperiodic ACF (AA-ACF) measurements, the PAPR performance and the PSD of the original and the IDFT-transformed signals. The "pilot-less" results from the baseline simulation are also included to illustrate the impact of the pilot-can-icr insertions, Figure 13 is a clear indication of the shift in the correlation properties of the transformed signal S after pilot insertion. l'he pilot-pattern insertion has a repetitive nature, for the PPI pilot pattern this is every 1 2 carriers, resulting in introducing periodic signals. The AA-ACF values are effectively halved.

This then translates into a halving of the PAPR reduction gains as shown in the CCDF curves of Figure 14. However, the 2 dB gains are still significant enough to make IJMT-transformation a valid solution particularly knowing that it holds zero-loss' of capacity attribute.

Finally, Figure 15 illustrates that the out-of-band characteristics remain unchanged.

Figures 16 and 17 demonstrate the relationship between the pilot-density per OFDM symbol and the PAPR performance: Figure 1 6 shows this in histovtm form, while Figure 17 shows the corresponding CCDF curves. Two exemplary pilot patterns PP1 and PP7 are plotted: with PPI representing a high-density pilot population (PP 1: Dx3, Dy=4 resulting in one in 12 pilot spacing per 1 5 OFDM symbol) & PP7 representing a low-density pilot population (PP7: Dx24, Dy4 resulting in 1 in 96 pilot spacing per OFDM symbol). It is evident that increased pilot-density reduces the PAPR reduction gain but. in fins ease, comparing pilot patterns PP1 and PP7, the gap is not that large.

In the embodiment illustrated in Figure 4A, QAM data corresponding to payload and pilots are passed through separate UMF-ftansformation and then mixed before entering the lEFT process. The performance evaluation revealed very small improvements. Nonetheless Figures 18 to 21 illustrate the results of the performance evaluation for this embodiment.

In simulating the embodiment of Figure 4A, the following simulation parameters were used (in fact these are the same as for the simulation of the Figure 3A embodiment): * FFT Size: * 32K Extended-Bandwidth Model * QAM:QPSK * Guard Interval: 1/4 * FEC: None * Pilot Pattern: PP1 [Dx3, Dy=44Dx*Dy 12] * Unitary Matrix Transformation: Inverse DFT * Ovcrsaiupling Ratio: 4 Figures 18 and 19 illustrate that there is little improvements in AA-ACF properties ia the embodiment of Figure 4A but this small improvement can only translate to a tiny enhancement of the PAPR perfonnance (in the order of 0.1-0.2 dB), over the perfonnance of the Figure 3A embodiment, as evident in the respective CCDF curves in Figures 20 and 21.

In the embodiments of Figure SA to 7A, the UMT pre-IFET process and pilot signal insertion of Figures 3A or 4A are extended. The IIMT-transformed PAPR reduction is a prc-IFFT method with zero-loss of complexity and reasonably small order of complexity. To determine whether the PAPR reductions might be thrthe.r increased by including a post-IFFI PAPR reduction algorithm, two known candidate post-lEFT techniques were considered as example schemes.

In each case, the common simulation parameters were set as below: FF1 Size: 32K * Bandwidth Extension: Enabled * QAM:16QAM * Guard Interval: 1⁄4 * FEC: None * Pilot Pattern: PP1 [Dx3, Dy4j * Pre-lEFT PAPR Reduction Method: Unitary Matrix Transformation (Inverse DFT) of Input Data IC) to lEFT Processor * Number of OFDM Symbols: 100,000 * Oversampling Ratio: 4 Simulation parameters specific to TR were set as below: 1 5 * Number of Tterations: 9 * Velip Values: o Without UIMT: 2.65 o WithUMT:2.5 Note: The Vclip value, when UMT is applied, is slightly lowered to ensure maximum number of iterations i.e. 9' is achieved for majority of OFDM symbols. This is because ITMT is expected to lower the maximum peak values of every OFDM symbol. This makes the comparison fair.

Simulation parameters specific to ACE were set as below: * Gain Value: 20 * Maximum Extension Value: 1.2 * Vclip Values: o Without UMT: 2.0dB o WithUMT:1.2dB Note: The Vclip value, when UMT is applied, is lowered to ensure that the same amount of clipping is applied. This is because UMT is expected to lower the maximum peak values of cvery OFDM symbol. This makes the comparison fair.

Figure 22 shows that the required dummy load insertions has significantly impacted the autocolTelation properties of the UMT-transfonned data (nearly undone what UMT had achieved). As a result, as the CCDF curves in Figure 23 show, the PAPR reductions arc mainly dominated by the TR algorithm with UMT only offering fractions of 1 dB. Typical values for the PAPR reductions in pilot-assisted OFDM systems based on a UMT alone are around 1.3dB, whilc with YR alone the same simulation offers a reduction of 1.6dB in combination however the reduction is 1.7dB.

The same is not true for the combination of IJMT and ACE. As Figure 24 illustrates, the autocorrelation properties of the signal feeding the lEFT have remained unchanged and thus a large gap appears between the AA-ACF values with and without UMT.

Figure 25 shows CCDF curves indicating the PAPR performance of pre-and post-JFFT schemes when ACE is used together with or in place of a UIMT pre-IFFT scheme. As anticipated, in this ease the PAPR reduction gains are cumulative resulting to a significant net gain. Typical values for the PAPR reductions in pilot-assisted OFDM systems based on a UIVIT alone are around 1.3dB, while with ACE alone the same simulation offers a reduction of 1.4dB -in combination the reduction is 2.7dB.

The embodiments of Figures 3A to 7A meet the majori of the conditions of an ideal' PAPI 1 0 reduction solution. There are no restrictions in deploying these schemes in any OFDM-based communication systems where excessive PAPR levels might he a critical issue. The results set out above relate to transmission systems modelled in Figures 3A to 7A that each include only scattered pilots.

Introducing continual pilots trirther degrades the benefit of the application of UMT prior to IFFT.

To illustrate this, the platform shown in Figure 3A was configured with the following simulation parameters (sometimes referred to as UK DVB-T2 mode): * FFT size: 321K * Extended Bandwidth Mode: Number of cariers2784l * QAMTypc:256 * Scattered Pilot (SP) Pattern: 7 [Dx=24, Dy=4] * SP Boosting Factor: 7/3 [7.4 dB] * Continual Pilot (CP) Patterns: 1 80 randomly positioned carriers as defmed in ETSI EN 302755, Vl.3.l * CP Boosting Factor: 8/3 [8.5 dli] * Guard Interval Fraction: 1/128 * Oversampling Ratio: 4 * Total Number of OFDM Symbols: 500,000 Only simulation results that are different to results shown in relation to Figures 8 to 12 are presented. The BER performance and the PSD characteristics remain unchanged (from Figures 10 and 12 respectively) and therefore not repeated here.

Figure 27 illustrates the PAPR Performance of Original and 1DFT-Transformed Signals with scattered pilots alone and with both scattered pilots and continual pilots. Comparison between Figure 27 and 9 (which relates to the application of a DM1 to pilot-less OFDM) reveals the fact that the PAPR reduction gain drops from 3.0 to 1.4dB (halving) when boosted scattered pilot (SP) carriers are inserted.

The PAPR reduction gains are then lowered to nearly zero dB when both scattered and continual pilots (CP) are inserted. Note that the curves for pilot-aided OFDM where no [IrviT' is applied are virtually identical whether scattered pilots are inserted alone or in combination with continual pilots; the respective curves therefore overlap as may he seen in Figure 27.

Figures 28A and 28B show how this degradation in the PAPR reduction gain occurs. Compare the pilot-less simulations in Figure 28A to the pilot-aided simulations (for different combinations of SP and CP) in Figure 28B, in each ease with and without U'MT. When SPs are inserted, the mean of absolute aperiodic ACF (aaACF) moves from 2.6x1 06 to 2.9x106. en both SPs and CPs are inserted, the two histograms are no longer separated by a significant amount but in fact overlapping.

As noted previously, nearly all OFDM-based transmission systems (including DVB-T2) rely on the presence of both pilot carrier types-Therefore, an extended solution is required if IJMT-based PAPR reduction is to he deployed in practical pilot-aided OFDM systems.

Figures 29 and 30 show the modified structures for the enhanced transmitter and receiver, respectively. The modified structure seeks to eliminate the issue raised in relation to the insertion of both SP and CP at the expense of increased complexity.

Figure 29 shows a schematic block diagram of a transmitter in an enhanced Pilot-Assisted UMT-OFDM System. Likewise, Figure 30 shows a schematic block diagram of a receiver for an enhanced Pilot-Assisted IJIMT-OFDM System.

In the transmit path of Figure 29, input bit streams 2920 are mapped to modulated data symbols (Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 2902. These modulated symbols are applied in parallel (having been converted from serial to parallel by a S/P converter 2904) to a unitary matrix transformer unit 2940 prior to transmission to an OFDM generation module 2910. In addition the modulated symbols are also applied to the OFDM generation module 2910 without being subjected to the unitary matrix transformer unit 2940 first.

At the transmitter, then, two different versions of the OFDM signal are generated. The first version is a conventional OFDM generation (without UMT transformation) -following the lower path in Figure 29: the second version differs in that LJMT transformation is applied to data QAIvI symbols (similar to the transmission path in Figure 3A), i.e. the tipper path in Figure 29. A Selection Process' 2950 is then applied to determine which OFDM symbol is to be transmitted. The selection criterion is conveniently one of choosing the version of OFDM symbol with the lower PAPR and transmitting a single bit 2952 to the receiver to indicate whether UMT transformation is applied or not for each respective OFDM symbol.

In this example, randomized boosted BPSK pilot data (output by a BPSK mapper 2932) is inserted by respective multiplexers 2942, 2942'.

In the ease of the generation of the first version of the OFDM signal, the pilot data is multiplexed with the parallelised, modulated data symbols by multiplexer 2942'. The resulting multiplexed signal is applied to the OFDM generation module 2910 without being subjected to the unitary matrix transfonner unit 2940.

in the case of the generation of the second version of the OFDM signal, the pilot data is multiplexed with the paralleliscd, modulated data symbols by multiplexer 2942. The resulting version of the multiplexed signal is applied to the OFDM generation module 2910 only after being subjected to the unitary matrix transformer unit 2940.

The OFDM generation module 2910 generates the second version of the OFDM signal by analogy with the transmit path in Figure 3A. The OFDM generation module 2910 includes an edge carrier insertion unit 2906, an Inverse Fast Fourier Transfer (IFFT) arithmetic operation unit 2908, and a guard interval (Cl) insertion unit 2912. Tn the OFDM generation module 2910, the modulated symbols arc allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 2908. The multiplexed symbols output by multiplexer 2942 are applied first to the edge carrier insertion unit 2906, and then to the Inverse Fast Fourier Transfer (WET) arithmetic operation unit 2908.

The OFDM generation module 2910 includes a further edge earner insertion unit 2906' and a farther Inverse Fast Fourier Transfer (WET) arithmetic operation unit 2908'. To generate the first version S of the OFDM signal, the multiplexed symbols output by multiplexer 2942' are applied first to the further edge carrier insertion unit 2906', and then to the further Inverse Fast Fourier Transfer (WFT) arithmetic operation unit 2908'.

The output signals from both IFFT arithmetic operation units 2908 and 2908' are received by the Sclcction Process 2950. The version of OFDM symbol determined to have the lower PAPR is then processed by the GI insertion unit 2912 so that an appropriate guard interval is provided. In example embodiments the selection process maybe configured to operate by circuitry, The structure of the enhanced receiver remains mostly the same as the low complexity receiver (i.e. that illustrated in Figure 3B) with the exception of an additional dc-multiplexer 3010. The receiver acts on the received signalling bit 2952 to bypass the inverse UIMT transformation or not.

1 5 In the enhanced receive path of Figure 30, the demodulation of the OFDM signal obtained through the modulatlon can be carried out by using a Fast Fourier Transfer (FF1) arithmetic operation 3054 for carlying out Fourier transfonn.

The receive path includes a guard interval removal unit 3052, Fast Fourier Transfer (EFT) arithmetic operation unit 3054, a channel estimator unit 3056, a channel equalizing unit 3058, an edge carrier removal unit 3060, a demuhiplexer 3066, an inverse unitary matrix transformer unit 3070, a parallel to serial converter 3062 and a. demapping unit 3064.

The signal output by the channel equalizing unit 3058 is applied to two branches: a first branch in which the inverse unitmy matrix transformer unit 3070 is bypassed and a second branch where that unit is applied. The additional dc-multiplexer 3010 receives the signalling bit 2952 corresponding to the signal output by the channel equalizing unit 3058 and determines which of the signals (i.e. first branch or second branch) is to be applied to the parallel to serial converter 3062.

The enhanced transmitter entails a degree of additional complexity. 1-lowever, the increase in complexity is considered negligible at the enhanced receiver. There is an additional consideration regarding the robust transmission of the I bit side information that has to he available at the receiver per OFDM symbol.

To illustrate the benefit of implementing an enhanced Pilot-Aided UMT-OFDM System, the system illustrated in Figures 29 and 30 were simulated. The simulation platform was constructed and configured according to the same simulation parameters (UK DVB-T2 mode) listed in relation to Figures 27, 28A and 28B.

Figure 31 shows the Complimentary Cumulative Distribution Function (CCDF) indicative of the PAPR performance for both simplified and enhanced structures when boosted SP and CP cariers are inserted.

The PAPR reduction increases from nearly zero (where only the Figure 3A transmitter path is available) to 1.2 dB (using the enhanced transmitter of Figure 29), justifying the increased complexity when inclusion of both sets of pilots is essential (such as DVB-T2).

As for the simplified system in relation to Figures 5A to 7B, die enhanced system can benefit from the application of post-WFT PAPR techniques. Both TR and ACE techniques are considered in detail in combination with the enhanced system of Figures 29 and 30.

[he proposed enhanced structure shown in Figure 29 is then modified to include one of the two S post-IFFT schemes, as illustrated in Figure 32. t'he receiver, for both cases, remains the same as shown in Figure 30, as TR and ACE algorithms are transmit only processes. For TR, some of the carriers are reserved and hence signalling is sent to the receiver to convey this information.

In the transmit path of Figure 32, input bit streams 3220 are mapped to modulated data symbols (Quadrature Amplitude Modulated (QAM) symbols) by a mapping unit 3202. These modulated symbols are applied in parallel (having been converted from serial to parallel by a S/P converter 3204) to a unitary matrix transformer unit 3240 prior to transmission to an OFDM generation module 3210. hi addition the modulated symbols are also applied to the OFDM generation module 3210 without being subjected to the unitary matrix transformer unit 3240 first, At the transmitter, then, two different versions of the OFDM signal are generated. The first I 5 version is a conventional OFDM generation (without LJMT transformation) -following the lower path in Figure 32: the second version differs in that UMT transformation is applied to data QAM symbols (similar to the transmission path in Figure 5A). i.e. the tipper path in Figure 32. A Selection Process' 3250 is then applied to determine which OFDM signal is to be transmitted. The selection criterion is conveniently one of choosing the version of OFDM signal with the lower PAPR and transmitting a single bit 3252 to the receiver to indicate whether UMT transformation is applied or not for each respective OFDM symbol.

In this example, randomized boosted BPSK pilot data (output by a BPSK mapper 3232) is inserted by respective multiplexers 3242, 3242'.

In the case of the generation of the first version of the OFDM signal. the pilot data is multiplexed with the parallelised, modulated data symbols by multiplexer 3242'. The resulting multiplexed signal is applied to the OFDM generation module 3210 without being subjected to the unitary matrix transfonner unit 3240.

In the ease of the generation of the second version of the OFDM signal, the pilot data is multiplexed with the parallelised, modulated data symbols by multiplexer 3242. The resulting version of the multiplexed signal is applied to the OFDM generation module 3210 only after being subjected to the unitary matrix transformer unit 3240.

The OFDM generation module 3210 generates the second version of the OFDM signal by analogy with the transmit path in Figure SA. The OM generation modLile 3210 includes an edge carrier insertion unit 3206, an Inverse Fast Fourier Transfer (lEFT) arithmetic operation unit 3208, and a guard interval (01) insertion unit 3212. In the OFDM generation module 3210, the modulated symbols arc allocated to plural sub-carriers, respectively, each of the symbols being transformed by the Inverse Fast Fourier Transfer (IFFT) arithmetic operation 3208. The multiplexed symbols output by multiplexer 3242 are applied first to the edge carrier insertion unit 3206, and then to the Inverse Fast Fourier Transfer (lEFT) arithmetic operation unit 3208.

The OFDM generation module 3210 includes a further edge carrier insertion unit 3206' and a further Inverse Fast Fourier Transfer (lEFT) arithmetic operation unit 3208'. To generate the first version of the OFDM signal, the multiplexed symbols output by multiplexer 3242' are applied first to the further edge carrier insertion unit 3206', and then to the further Inverse Fast Fourier Transfer (lEFT) arithmetic operation unit 3208'.

The output signals from both IFFI' arithmetic operation units 3208 and 3208' are received by the Selection Process 3250. The version of OFDM symbol determined to have the lower PAT'R is then S applied to a post-IFFT PAPR reduction process 3215. One the (or each) post-lEFT PAPR reduction 1; process has been applied the output signals from this process are then processed by the GE insertion unit 3212 so that an appropriate guard interval is provided.

Figure 33 shows the PAPR performance of the combined pre-and post-IFFT processes when the post-1FFF PAPR technique is Tone Reservation (TB). Here the simulation parameters remain those for 1 0 uK DVB-T2 mode, with the following additional parameters to configure the TR algorithm: o V111, 2.65 when UMT disabled) o = 2.5 (when UMT enabled) o Number of Iterations = 10 Note that the Yclip is slightly lowered when UJvIT is enabled. This is simply because the pre- lEFT' (UMT transfonnation) on average would have reduced the peak power values and hence the post-lIFT (TR) should be adjusted accordingly.

These results of the simulation illustrated in Figure 33 are summarized as follows: PAPR Reduction Gain (dB) in (SP & CP) pilot-assisted OFDM systems based on a UMT alone are around 1.3dB, while with TR alone the enhanced simulation offers a reduction of 1.2dB -in combination however the reduction is 2.0 dB. Thus either the prc-or the post-IFFT schemes on their own would provide the same PAIPR reduction gains (around 1.25 dB) bitt when combined this increases to 2.0 dB.

Figure 34 shows the PAPR perfonnance of the combined pre-and post-lEFT processes when the post-[FFT PAPK technique is Active Constellation Extension (ACE). Here the simulation parameters remain those for UK DVB-T2 mode, with the following additional parameters to configure the ACE algorithm: o V111, 2.0 dB (when UMI disabled) o V = 1.2 dB (when IJMl' enabled) o Gain'20 o LValue=1.2 The two values differ for the same reasons explained above.

These results of the simulation illustrated in Figure 34 are summarized as follows: PAPR Reduction Gain (dB) in (SP & Cl') pilot-assisted OFDM systems based on a UMT alone are around 1.3dB, while with ACE alone the enhanced simulation offers a reduction of just 0.6dB -in combination however the reduction is 1.8 dB.

For this particular configuration. the PAPR reduction gains are nearly cumulative, resulting in 1.8 dli when both UMT and ACE (pre-and post-WET PAPR reductions) are turned on.

Any new post-lEFT PAPR reduction algorithm that may be applied whilst respecting the UMT-transformation conditions may be expected to providc independent high PAPR reduction gains with reduced complexity.

The following numbered clauses provide further exemplary aspects and features of the present technique: I. A transmitter for transmitting data in a wireless communications sysl.em, the transmitter comprising: a converter having circuitry configured to convert input modulated symbols into respective data vectors; S a unitary matrix transformer unit having circuitry configured to transform the data vectors into corresponding transformed vectors; a multiplexer having circuitry configured to receive input modulated pilot symbols and the transformed vectors and to output a plurality of multiplexed transformed vectors; and an Orthogonal Frequency Division Multiplexing, OFDM, symbol generator having circuitry configured to generate OFDM symbols from the multiplexed transformed vectors, each OFDM symbol including a plurality of sub-carrier symbols formed in the frequency domain and transformed by inverse fast Fourier transform, IFFT, into a plurality of samples in the tune domain for transmission.

2. A transmitter as claimed in clause 1, wherein the multiplexer inserts the pilot symbols amongst the multiplexed transformed vectors.

3. A transmitter as claimed in clause 1, wherein the input modulated pilot symbols are received as a pilot data vector, the pilot data vector having been transformed according to a unitary matrix transformation.

4. A transmitter as claimed in clause 1, wherein the multiplexer further receives dummy data, and the dummy data is multiplexed with the transformed data vector and the received modulated pilot symbols; the transmitter further comprising: a post-IFFT processor which operates to process OFDM symbols after IFFT to calculate the positions of peak power amplitude in the time domain; and to cancel detected peaks in the generated OFDM symbols in accordance with a predeternnned peak cancellation signal generated from the dummy data.

5. A transmitter as claimed in clause I, wherein the transmitter further comprises a post-IF FT processor which operates to process OFDM symbols after lEFT by extending outer constellation points in the frequency domain; thereby modifying the power distribution of the OFDM symbols in the time domain.

6. A method for transmitting data in a wireless communications system, the method comprising: converting input modulated symbols into respective data vectors; transforming by circuitry the data vectors into corresponding transformed vector according to a unitary matrix fran sformati on; receiving input modulated pilot symbols; multiplexing the transformed vector with information corresponding to the received modulated pilot symbols to output a plurality of multiplexed transformed vectors; and gen crating by circuitry OFDM symbols from the multiplexed tran sfonn ed vectors, each OF DM symbol including a plurality of sub-carrier symbols fonned in the frequency domain and transformed by inverse fast Fourier transform. WET, into a plurality of samples in the time domain for transmission.

7. A method as claimed in clause 6, wherein the multiplexing step scatters the pilot symbols amongst the transformed data vectors.

8. A method as claimed in clause 6, wherein the step of receiving input modulated pilot symbols comprises converting input modulated pilot symbols into a pilot data vector; and transforming the pilot data vector according to a unitary matrix transformation.

9. A method as claimed in clause 6, wherein the multiplexing step further includes receiving dummy data, and multiplexing the dummy data with the transformed data vector and the received modulated pilot symbols; the method further comprising: processing OFDM symbols after lEFT to calculate the positions of peak power amplitude in the time domain; and cancelling detected peaks in the generated OFDM symbols in accordance with a predetermined peak cancellation signal generated from the dummy data..

10. A method as claimed in clause 6, the method further comprising: processing OFDM symbols after WET by extending outer constellation points in the frequency domain; thereby modizing the power distribution of the OFDM symbols in the time domain.

11. A receiver for receiving data in a wireless communications system, the receiver comprising: a demodulator having circuitry configured to demodulate Orthogonal Frequency Division Multiplexing, OFDM, symbols incorporating modulated pilot signals, the demodulation being carried out by using a fast Fourier transform, FFT. and to generate a plurality of multiplexed transformed vectors, each O.EDM symbol including a plurality of sub-carrier symbols formed in the frequency domain; a demultiplexer having circuitiy configured to receive a plurality of multiplexed transformed vectors and to output modulated pilot symbols and transformed vectors; a channel estimator having circuitry configured to receive said pilot symbols and to synchronise the channels in accordance with the pilot signals; and an inverse unitary matrix transformer unit having circuitry configured to transform transformed vectors into respective data vectors corresponding to a received data hit stream.

12. A receiver as claimed in clause 11, wherein the demultiplexer extracts the pilot symbols from amongst the multiplexed transfonned vectors.

13. A receiver as claimed in clause 11, wherein the received pilot symbols are transformed according to an inverse unitary matrix transformation before transmission to the channel estimator.

14. A method for receiving data in a wireless communications system, the method comprising: demodulating by circuitry Orthogonal Frequency Division Multiplexing. OFDM, symbols incorporating modulated pilot signals. the demodulation being carried out by using a fast Fourier transform, FFT, to generate a plurality of multiplexed transformed vectors, each O.EDM symbol including a plurality of sub-carrier symbols formed in the frequency domain; demultiplexing a plurality of multiplexed transfonncd vectors into output modulated pilot symbols and transformed vectors; synchronising channels in accordance with the pilot signals; and transforming by circuitry transformed vectors into respective data vectors corresponding to a received data bit stream in accordance an inverse unitary matrix transformuation.

15. A method as claimed in clause 14, further comprising: transforming the received pilot symbols according to an inverse unitary matrix transformation before the chaiinel synchronisation step.

Various further aspects and features of the present disclosure are defmed in the appended claims.

Further example aspects and features of the present disclosure are defined in the appended claims.

Various combinations of features may be made of the features and method steps defmed in the dependent claims othei than the specific combinations set out in the attached claim dependency. Thus the claim dependencies should not he taken as limiting.

Claims (15)

1. CLAIMS1. A transmitter for transmitting data in a wireless communications system, the transmitter comprising: a converter having circuitry configured to convert input modulated symbols into respective data vectors; a unitary matrix transformer unit having circuitry configured to transform the data vectors into coiTespondrng transformed vectors; a multiplexer having circuitry configured to receive input modulated pilot symbols and the transformed vectors and to output a plurality of multiplexed transformed vectors; and an Orthogonal Frequency Division Multiplexing, OFDM, symbol generator having circuitry configured to generate an OFDM symbol from the multiplexed transformed vectors, the OFDM symbol including a plurality of sub-carrier symbols formed in the frequency domain and transformed by inverse fast Fourier transform, lEFT, into a plurality of samples in the time domain for transmission.
2. 2. A transmitter as claimed in Claim 1, wherein the multiplexer inserts the pilot symbols amongst the multiplexed transformed vectors.
3. 3. A transmitter as claimed in Claim 1, wherein the input modulated pilot symbols are received as a pilot data vector, the pilot data vector having been transformed according to a unitary matrix transformation.
4. 4. A transmitter as claimed in Claim 1, wherein the multiplexer further receives dummy data, and the dummy data is multiplexed with the transformed data vector and the received modulated pilot symbols; the transmitter further comprising: a post-IFFT processor which operates to process OFDM symbols after IFFT to calculate thc positions of peak power amplitude in the timc domain; and to cancel detected peaks in the generated OFDM symbols in accordance with a predetennined peak cancellation signal generated from the dummy data.
5. 5. A transmitter as claimed in Claim 1, wherein the transmitter further comprises a post-IFFT processor which operates to process OFDM symbols after IFFT by extending outer constellation points hi the frequency domain; thereby inodizing the power distribution of the OFDM symbols in the timnnc domain.
6. 6. A method for transmitting data in a wireless communications system, the method comprising: converting input modulated symbols into respective data vectors; transforming by circuitry the data vectors into corresponding transformed vector according to a unitary matrix transformation; receiving input modulated pilot symbols; multiplexing the transformed vector with infonnation corresponding to the received modulated pilot symbols to output a plurality of multiplexed transformed vectors; aud H gciicratng by circuitry an OFDM symbol from the multiplexcd transformed vectors, the OFDM symbol including a plurality of sub-carrier symbols formed in the frequency domain and transformed by inverse fast Fourier transform. IFFT, into a plurality of samples in the time domain for transmission.H
7. 7. A method as claimed in Claim 6, wherein the multiplexing step scatters the pilot symbols amongst the transformed data vectors.
8. 8. A method as claimed in Claim 6, wherein the step of receiving input modulated pilot symbols comprises converting input modulated pilot symbols into a pilot data vector; and transforming the pilot data vector according to a unitary matrix transformation.
9. 9. A method as claimed in Claim 6, wherein the multiplexing step further includes receiving dummy data, and multiplexing the dummy data with the transformed data vector and the received modulated pilot symbols; the method further comprising: processing OFDM symbols after IFFT to calculate the positions of peak power amplitude in the H time domain; and cancelling detected peaks in the generated OFDM symbols in accordance with a predetermined 1; peak cancellation signal generated from the dummy data.
10. 10. A method as claimed in Claim 6, the method further comprising: processing OFDM symbols after IFFI' by extending outer constellation points in the frequency domain; thereby modifying the power distribution of the OFDM symbols in the time domain.
11. 11. .A receiver for receiving data in a wireless communications system, the receiver comprising: a demodulator having circuitry configured to demodulate Orthogonal Frequency Division Multiplexing, OFDM, symbols incorporating modulated pilot signals, the demodulation being calTied out by using a fast Fourier ti'anslbrm, FFT, and to generate a plurality of multiplexed transformed vectors, each OFDM symbol including a plurality of sub-carrier symbols formed in the frequency domain; a demultiplexer having circuitry configured to receive a plurality of multiplexed transformed vectors and to output modulated pilot symbols and transformed vectors; a channel estimator having circuitry configured to receive said pilot symbols and to synchronise the channels in accordance with the pilot signals; and an inverse unitary matrix transfonner unit having circuitry configured to transform transformed vectors into respective data vectors corrcsponding to a received data bit stream.
12. 12. A rcccivcr as claimed in Claim 11, wherein the demultiplexer extracts the pilot symbols from amongst the multiplexed transformed vectors.
13. 13. A receiver as claimed in Claim 11, wherein the received pilot symbols are transformed according to an inverse unitary matrix transformation before transmission to the channel estimator.
14. 14. A method for receiving data in a wireless coimnunications system, the method comprising: demodulating by circuitry Orthogonal Frequency Division Multiplexing, OFDM, symbols incorporating tnodulatcd pilot signals, the demodulation being carried out by using a fast Fourier transform. FFT. to generate a plurality of multiplcxcd transformed vectors, each OFDM symbol including a plurality of sub-carrier symbols formed in the frequency domain; deinultiplexing a plurality of multiplcxcd transformed vectors into output modulated pilot symbols and transformed vectors; synchronising channcls in accordance with the pilot signals; and transforming by circuitry transformed vectors into respective data vectors corresponding to a received data bit stream in accordance an inverse unitary matrix transformation.
15. 15. A method as claimed in Claim 14, further comprising: transfonning the received pilot symbols according to an inverse unitary matrix transformation before the channel synchronisation step.