Blind decision feedback equalizer using dual-variance Gaussian clustering algorithm for wireless communications

Intersymbol interference (ISI) of the radio propagation through multipath fading channels is a major factor that limits the performance of mobile communication systems. Blind decision feedback equalization (DFE) is one of major methods for channel equalization. We propose a new method, based on a Hammerstein filter instead of DFE filters, for enhancing the performance of DFE, hence, blind HDFE. In this paper, we show the bit error rate (BER) performance of blind DFE and blind HDFE, in order to prove the superiority of BHDFE.

Computationally efficient methods for blind decision feedback equalization of QAM signals

The addressed blind decision feedback equalizer (DFE) reverses the classical order of its feed-forward and feedback filters at the beginning of adaptation to achieve the best equalization of the minimum and maximum phase components of a channel transfer function. Although very effective, this blind equalization approach deals with the feedback filter mismatch at the time of its transformation from the front-end all-pole whitener of the received signal to the decision-directed feedback equalizer placed after the feedforward filter. To eliminate this weakness, the adaptive neuron slope is introduced instead of the fixed one into the stochastic gradient whitening algorithm based on the joint entropy maximization cost. The performance of the innovated algorithm is verified by simulating m-QAM (quadrature amplitude modulation) signals transmission over multipath channels. The algorithm with the adaptive neuron slope achieves a better whitening of the received signal spectrum, and, hence, increases the equalization successfulness.

For PC DRAM buses, the number of slots per channel has decreased as data rates have increased. This limits the maximum memory capacity per channel. Signal equalization can be used to increase bit-rates for channels with a large number of slots and offer a cost effective method to solve the memory capacity problem. This paper presents a blind adaptive decision feedback equalizer (DFE) that enables high data-rates with a large number of slots per channel. Measurements at 2.6 Gb/s over a four-drop bus are presented.

An important challenge for implementing optical signal processing functions such as wavelength conversion or wavelength data exchange (WDE) is to avoid the introduction of linear and nonlinear phase noise in the subsystem. This is particularly important for phase noise sensitive, high-order quadrature-amplitude modulation (QAM) signals. In this paper, we propose and experimentally demonstrate an optical data exchange scheme through cascaded 2nd-order nonlinearities in periodically-poled lithium niobate (PPLN) waveguides using coherent pumping. The proposed coherent pumping scheme enables noise from the coherent pumps to be cancelled out in the swapped data after WDE, even with broad linewidth distributed feedback (DFB) pump lasers. Hence, this scheme allows phase noise tolerant processing functions, enabling the low-cost implementation of WDE for high-order QAM signals. We experimentally demonstrate WDEs between 10-Gbaud 4QAM (4QAM) signal and 12.5-Gbaud 4QAM (16QAM) signal with 3.5-MHz linewidth DFB pump lasers and 50-GHz channel spacing. Error-free operation is observed for the swapped QAM signals with coherent DFB pumping whilst use of free-running DFB pumps leads to visible error floors and unrecoverable phase errors. The phase noise cancellation in the coherent pump scheme is further confirmed by study of the recovered carrier phase of the converted signals. In addition to pump phase noise, the influence of crosstalk caused by the finite extinction ratio in WDE is also experimentally investigated for the swapped QAM signals.

We present novel blind decision feedback equalization algorithms for VSB-based DTV receivers by modifying a stop-and-go dual-mode constant modulus algorithm (SAG DMCMA) and applying a selective feedback (SF) scheme to stop-and-go, (SAG) based algorithms. The modification of the SAG DMCMA applies the SAG scheme only to the feedback filter tap update of the DFE, thus reducing the loss of information caused by the SAG policy. The SF scheme, in which we choose the equalizer output or the slicer output for the input of the feedback filter in the DFE according to the SAG flag, diminishes error propagation and thus improves convergence performance. Considering VSB modulation and RF-band effects with the baseband-equivalent VSB channel model in our simulations raises the reliability of our results for practical applications.

The constant modulus algorithm (CMA) for blind equalization requires a separate carrier-recovery system for phase recovery. A modified CMA, called the multimodulus algorithm (MMA), which may perform joint blind equalization and carrier recovery without the need for a separate carrier-recovery system for quadrature amplitude modulation (QAM) signal constellations. This letter mathematically analyzes the steady-state mean square error (MSE) of MMA. Analysis results indicate that MMA produces 50% fewer steady-state MSE than CMA.

In this paper, a new blind equalization algorithm is proposed based on guard interval (GI)-restoral property for quadrature amplitude modulation (QAM) signals. The proposed algorithm utilizes existing redundancy in the time domain to completely eliminates inter-symbol-interference. We prove that the usage of GI-restoral property is sufficient to obtain a zero-forcing equalizer of the channel. The main advantage of this approach resides in its simplicity as the equalizer coefficients are estimated using a second-order least square fitting (LSF) scheme. The equalizer uses the LSF in the initial phase and then switches to a proposed constant modulus algorithm (CMA). The proposed CMA is constructed through the exploitation of a fully data information. To achieve a dramatic improvement in equalization performance, a modified CMA (MCMA) is proposed for OFDM systems based on cyclic prefixing approach. The resulting scheme provides a state-of-the-art low complexity with superior performance blind equalization. Extensive simulation shows that LSF-MCMA has a better steady-state equalization performance than LSF-CMA.

Probabilistically shaped (PS) high-order quadrature amplitude modulation (QAM) signals are attractive to coherent optical communication due to increased spectral efficiency. However, standard digital signal processing algorithms are not optimal to demodulate PS high-order QAM signals. Therefore, a compromise equalization is indispensable to compensate the residual distortion. Meanwhile, the performance of conventional blind equalization highly depends on the accurate amplitude radius and distribution of the signals. The PS high-order QAM signals make the issue worsen because of indistinct amplitude distributions. In this work, we proposed an optimized blind equalization by utilizing a peak-density K-means clustering algorithm to accurately track the amplitude radius and distribution. We experimentally demonstrated the proposed method in a PS 256-QAM coherent optical transmission system and achieved approximately 1 dB optical signal-to-noise ratio improvement at the bit error rate of 1×10-3.

We propose and numerically investigate a method for direct in-phase quadrature (IQ) modulation of spin-controlled vertical-cavity surface-emitting lasers (spin-VCSELs) for coherent communications. In particular, an injection-locked spin-VCSEL is investigated by using spin-flip rate equations for generating quadrature amplitude modulation (QAM) signals. A polarizer 45° tilted to the polarization axis of the injected external light is used to extract the in-phase and quadrature-phase components of the output light from the injection-locked spin-VCSEL. It is confirmed that QAM signals can be generated with the proposed method successfully, and long electron spin relaxation time and small linear birefringence in the spin-VCSEL are suitable for efficiently generating clear QAM constellations with a low error vector magnitude. A detuning condition for the injection locking of the spin-VCSEL is found to be important for obtaining wide modulation bandwidth of the direct IQ modulation.

In this work, we base on multi-layered perceptron neural networks with backpropagation algorithm (MLP/BP) to construct soft decision feedback equalizers (DFEs). The proposal is used to recover distorted 16-point quadrature amplitude modulation (16-QAM) signal. For better performance, error control codes (ECC) are applied to enhance the accuracy of the transmitted data. From the simulations, we note that the MLP/BP-based soft DFEs with bit-interleaved TCM can recover severe distorted 16-QAM data as well as suppress intersymbol interference (ISI) and background additive white Gaussian noise (AWGN). As compared with the LMS DFE, the proposed scheme can provide better bit-error-rate (BER) and packet-error-rate (PER) performance.

Summary form only given. This paper presents a Variable Step Size (VSS) technique that is integrated with a Stop-And-Go (SAG) approach for blind adaptive equalization of Quadrature Amplitude Modulation (QAM) signals. The proposed technique is applied to the Constant-Modulus Algorithm (CMA) and the Multi-Modulus Algorithm (MMA). The SAG approach allows the VSS technique to either vary the step size or to hold the previous one depending on a specific error criterion. This leads to the reduction of the amount of computations involved Simulation results show that algorithms with the VSS technique result in less steady-state Mean-Squared Errors (MSEs) than those obtained for algorithms without the VSS technique for the same speed of convergence.

A dual‐mode multi‐modulus algorithm (DM‐MMA) and a stop‐and‐go dual‐mode multi‐modulus algorithm (SAG‐DMMMA) for blind equalisation of high‐order quadrature amplitude modulation (QAM) signals are proposed. Simulation results show the proposed blind equalisation algorithms have faster convergence speed and smaller steady‐state mean square error, compared with the recently introduced multi‐modulus algorithm. Copyright © 2008 John Wiley & Sons, Ltd.

Remote monitoring of patients using wireless capabilities increasingly plays an important role in healthcare and the communication with implanted medical devices (IMDs). In the case of IMDs, transmission is characterized by low peak power and low duty cycle to reduce the potential for adverse bio-effects. Currently, most IMDs use radio-frequency (RF) electromagnetic waves to communicate through the body. The corresponding maximum bandwidth allowed is 300 kHz as regulated by the FCC, which inherently limits the communication rates of these devices (limiting to maximum data rates of 50 kb/s). Beyond bandwidth restrictions, the main limitation for using RF electromagnetic waves in the body is loss of signal because of attenuation in the body. For these reasons the output power of RF devices is limited to 25 µW, which also limits data rates to 50 kb/s. In this study, an ultrasonic communication channel is explored for IMD communications as an alternative to EM transmission. We demonstrate the ability to communicate through tissues using the ultrasonic communications channel while maintaining data rates necessary for streaming of high definition video. In our experiments we used a matched pair of 5-MHz f/3 single-element transducers operating in a pitch-catch configuration in a tank of degassed water. Between the transducers was placed either a slab of pork tenderloin or a slab of beef liver (each more than 2 cm thick). Quadrature amplitude modulation (QAM) and quadrature phase shift keying (QPSK) signals were coded for transmission into the transmitting transducer using an arbitrary waveform generator. Signals were transmitted and captured by sending 10 snapshots (packets) of 50,000 samples (10,000 training / 40,000 decision-directed) using a fractionally-spaced (2 samples per symbol) decision feedback equalizer with up to 40 taps in the feed forward section, and 40 taps in the feedback section. Signals were successfully decoded through the samples using QPSK, 16QAM and 64QAM codes with bit error rates of (BER) < 1e-4. A maximum data rate of 30 Mbps using the 64QAM was achieved through both the pork tenderloin and the beef livers. These experiments successfully demonstrate that communications through the ultrasonic channel can provide data rates capable of streaming high definition video.

In this paper, we focus on the automatic modulation for unsynchronized quadrature amplitude modulation (QAM) signals in both time and frequency in the practical communication systems. In practice, the receiver has little prior knowledge about the transmitted signal such as the timing error and frequency offset, directly applying the log-likelihood estimation to recognize the QAM signal would cause the decline in the recognition performance. To resolve this challenging problem, the paper presents the improved recognition scheme: before recognition, apply the Gardner loop timing synchronization and differential approach to the received signal. Simulation result shows that the proposed scheme is robust to timing and frequency offset and when SNR≥12dB and the number of symbols N is more than 1000, the successful classification rate of the 16QAM signal reach 98%.