^{1}

Physical layers of communication systems using Filter Bank Multicarrier (FBMC) as a modulation scheme provide low out-of-band leakage but suffer from the large Peak-to-Average Power Ratio (PAPR) of the transmitted signal. Two special FBMC schemes are investigated in this paper: the Orthogonal Frequency Division Multiplexing (OFDM) and the Staggered Multitone (SMT). To reduce the PAPR of the signal, time domain clipping is applied in both schemes. If the clipping is not compensated, the system performance is severely affected. To avoid this degradation, an iterative noise cancelation technique, Bussgang Noise Cancelation (BNC), is applied in the receiver. It is shown that clipping can be a good means for reducing the PAPR, especially for the SMT scheme. A novel modified BNC receiver is presented for SMT. It is shown how this technique can be implemented in real-life applications where special requirements must be met regarding the spectral characteristics of the transmitted signal.

In wireless communications the frequency spectrum is an essential resource. As the unlicensed spectrum is used by an increasing number of devices, the possibility of communication collision is increasing. To avoid this collision, two solutions are possible: extending the frequency limits higher to unused frequency bands at the upper end of the spectrum or reaggregating the densely used licensed frequency bands. Both ideas have disadvantages: the use of higher frequencies requires expensive specially designed analog devices; the reuse of the spectrum calls for complex, intelligent, and adaptive systems. In this paper the focus is on the reuse of the spectrum with multicarrier modulations tailored for spectrally efficient applications.

Future applications operating in the licensed bands, for example, cognitive radios, favor spectrally efficient FBMC schemes with low out-of-band leakage, minimizing harmful interference between devices using adjacent channels. In this paper two subclasses of FBMC are investigated, both allowing the use of a complex modulation alphabet: OFDM and SMT. Both of these schemes provide relatively low out-of-band leakage.

Today OFDM [

Numerous signal processing methods have been proposed to reduce the PAPR of OFDM [

Besides OFDM, another FBMC-based multicarrier family is being strongly investigated: the SMT [

In this paper the baseband amplitude clipping method [

Decision Aided Reconstruction (DAR) [

Bussgang Noise Cancellation (BNC) [

This paper is organized as follows. First the system model applying clipping on coded FBMC signals is introduced. A short description is given of OFDM and SMT. The PAPRs and the spectral characteristics of the transmitted signals are compared. The mathematical description of the clipping effects is also presented. In the next section, a detailed description is given of the soft BNC receiver algorithm, introducing modifications to the method presented in [

The FBMC transmit signal is constructed from

Block diagram of an FBMC modulation scheme.

The OFDM scheme is a special class of FBMC where a prototype filter with a rectangular impulse response is applied. This leads to a simplified structure where the consecutive symbols do not overlap, that is,

Block diagram of an OFDM transmitter.

In the SMT scheme prototype filters with overlapping impulse responses fulfilling the Nyquist criterion are applied. Due to the advantageous properties of the prototype filter bank, the SMT signal will have a better Adjacent Channel Leakage Ratio (ACLR) than OFDM. With the use of offset-QAM modulation, where the real and imaginary data are transmitted with a time offset of a half symbol duration, no data rate loss will occur compared to OFDM. Prior to transmission, the symbols are overlapped such that they can be separated at the receiver. In order to maintain orthogonality of the filter bank structure, CP can not be used in SMT systems. As a result, techniques with higher complexity must be applied in comparison to OFDM in order to combat the channel-induced intersymbol interference [

Block diagram of an SMT transmitter.

The PAPR is one of the quantities that describes the dynamic properties of the transmitted signal

CCDF of the PAPR of the transmitted signal of OFDM and SMT for various number of subchannels. The probability that an amplitude value exceeds a certain threshold

Spectral behavior especially regarding the ACLR is also an important property of the transmitted signal. The power spectrum density function of the transmitted signal with an oversampling factor of 4 is depicted in Figure

Power spectrum density comparison of the transmitted signals of OFDM and SMT with various number of subchannels.

Clipping is applied to the baseband transmit signal

For modeling the transceiver chain the digital baseband equivalent is used. The model of the transceiver is presented in Figure

Block diagram of the baseband transceiver chain.

In OFDM systems, the CP of

Block diagram of the Bussgang noise cancelation for OFDM.

The extrinsic Log-Likelihood Ratio (LLR) for each channel observation

After interleaving the extrinsic LLRs provided by the channel decoder, the soft symbols are calculated as [

Each symbol is first weighted by the probability of the mapped bits and then summed up. Using these soft symbols a time domain estimation of the OFDM signal is performed. Clipping is applied with a level of

Subtracting the attenuated symbols from the clipped symbols, the estimated clipping noise can be expressed as

The BCJR channel decoder [

The convergence behavior of a turbo loop can be examined using the Extrinsic Information Transfer (EXIT) chart, developed by ten Brink [

The LLRs defined by (

The EXIT function of the BNC detector is not only a function of the a priori mutual information

EXIT chart, with iteration trajectories of the BNC turbo receiver with an

For SMT scheme the blocks of the BNC receiver presented for OFDM in Figure

The compensation of the clipping noise is performed in time domain before demodulation.

In the presence of ISI only a quasi Maximum Likelihood (ML) detection of the received modulation symbols

The demapping blocks have to be extended with additional signal processing blocks.

Block diagram of the modified BNC receiver for SMT.

First, an enlarged FFT operation is applied [

In real-life systems not all subcarriers are used for data transmission. Usually the DC subcarrier and some carriers at the edge of the transmission band are not used due to technical difficulties and guard band purposes in the spectrum. Clipping introduces nonlinear distortions in the entire baseband, so the originally unused subcarriers will contain components introduced by clipping. This also negatively affects the spectral behavior of the transmission signal, that is, leakage will appear. These components have to be suppressed. Digital filtering is not sufficient to suppress the clipping components on the unused subcarriers and analog filtering introduces modulation errors. Instead of filtering, the clipped transmit signal is demodulated again. The modulation values for each symbols of the used subcarriers are selected and the unused subcarriers are set to zero, and repeated modulation is performed similar as described in [

Block diagram of the modified transmitter of OFDM (a) and SMT (b) systems.

OFDM

SMT

The results of clipping and resetting the unmodulated subcarriers to zero for OFDM and SMT can be seen in Figures

CCDF of the PAPR values of the transmitted signal with clipping and additional signal processing for OFDM (a) and SMT (b).

OFDM

SMT

Power spectrum density function of the transmitted signal with clipping and additional signal processing for OFDM (a) and SMT (b).

OFDM

SMT

Table

Simulation parameters for SMT and OFDM system.

Parameter | SMT | OFDM |

Bandwidth | 8 MHz | |

Cyclic prefix ( |
0 | 128 |

Available subcarriers/subbands ( |
1024 | |

Modulated subcarriers/subbands |
768 | |

Overlapping factor ( |
4 | 1 |

Mapping ( |
4 (16-QAM) | |

Clipping ratio | 1 dB |

The binary data are encoded with a code rate of 1/2, using a 4-state recursive systematic convolutional encoder with polynomials

To obtain comparable bit error rates, the SNR normalized to one bit energy is defined. The noise power of the AWGN channel is calculated according to the following definition:

Excess delay and relative amplitude for IEEE 802.22 B and C channel profiles.

Profile B | Path 1 | Path 2 | Path 3 | Path 4 | Path 5 | Path 6 |
---|---|---|---|---|---|---|

Excess d. | −3 |
0 |
2 |
4 |
7 |
11 |

Rel. amp. | −6 dB | 0 dB | −7 dB | −22 dB | −16 dB | −20 dB |

| ||||||

Profile C | Path 1 | Path 2 | Path 3 | Path 4 | Path 5 | Path 6 |

| ||||||

Excess d. | −2 |
0 |
5 |
16 |
24 |
33 |

Rel. amp. | −9 dB | 0 dB | −19 dB | −14 dB | −24 dB | −16 dB |

The simulated BERs over AWGN channel can be seen in Figure

Bit error rates of the BNC receiver for OFDM and SMT signaling over AWGN channel.

The BER simulations for Channel B can be seen in Figure

Bit error rates of the BNC receiver for OFDM and SMT signaling over channel B.

The BER simulations for Channel C are shown in Figure

Bit error rates of the BNC receiver for OFDM and SMT signaling over channel C.

In this paper a modified BNC structure suitable for clipped SMT signal processing was presented. Based on the EXIT chart, it was shown that the proposed iterative scheme is convergent. It was also described how the clipping technique can be applied in real-life systems for both OFDM and SMT modulation. Finally, the performance of the BNC SMT receiver was verified and compared to OFDM based on BER simulations over AWGN and Rayleigh channels. For both systems the clipping compensation can be performed and the performance without clipping can be approached.

The research leading to these results was derived from the European Community’s Seventh Framework Programme (FP7) under Grant Agreement no. 248454 (QoSMOS).