An Adaptive Decision Feedback Equalisation For Broadband Powerline Communication System

Power line networks can be used to increase accessibility of broadband communication services in developing countries. Nevertheless, power line networks are affected by stochastic channel alterations triggered by load connection and disconnection, branched line lengths, branches, etc. This impairment affects the implementation of Broadband Power Line Communication (BPLC) system. This paper therefore proposes an Adaptive Decision Feedback Equalisation (ADFE) technique to overcome the stochastic channel changes in powerline communication channels. An appropriate power-line channel model is selected and channel impulse responses are obtained from the selected channel model. The impulse responses are obtained and used for simulation to analysing the the performance of ADFE technique. The ADFE is simulated and then results are analyzed through comparisons with other equalizers in order to examine its performance. Simulation results prove that the adaptive decision feedback equalizer performs better to overcome the effects of stochastic changes in power-line network compared to other techniques.

In response to this; Researchers Jones and Sistanizadeh presented channel equalization technique for very high speed digital data communication [4]. The performance of the equalizer was based on the application of decision feedback equalizer (DFE) which used a fractional space equalizer (FSE). The advantage of FSE was dominance performance that it offers for an arbitrary slicer sampling phase. Drajic and others proposed adaptive channel equalization technique to suppress multiple access interference (MAI) problems in wideband code division multiple accesses (WCDMA) downlink [5]. In this method, the conventional rake receiver is first defined followed by chip level equalizer known as common pilot channel.
Rice and Satorius proposed equalisation technique which was intended for multipath mitigation in flight telemetry [6]. Knowledge-Based Adaptive Channel Equalization is another method proposed by Huihua and others [7]. This method envisioned to resolve the problem of inter-symbol interference (ISI) in gradually varying channels. In 2002, Singer and others studied about turbo equalization approach to coded data transmission over channels with intersymbol interference [8]. They presented algorithms for turbo equalization, which are applied to coded data transmission over ISI channels. Eynard and Laot presented a blind Self -Adaptive Decision Feedback Equalizer (SA-DFE). The equalizer switches to a linear recursive equalizer when the channel is very dispersive and to a decision-directed DFE when the output of the equalizer presents reliable decisions. So their study was intended on giving details about the two structures of the SA-DFE and justifies why these structures are relevant.
Degardin and others proposed a semi-blind equalization algorithm, the algorithm associated with channel state change detection to handle variations in the transfer function [9]. The algorithm was designed to suit transmission on indoor power line. This approach serves as extension of blind estimation. Chuah (2007) proposed Adaptive Robust Turbo Equalization for PLC which acts as inner decoder and channel decoder acts as the outer decoder and it is called turbo equalizer [10]. The idea of turbo equalization lies in the exchange of soft information between the equalizer and the decoder. The turbo equalizer needs knowledge of Channel Impulse Response (CIR) to perform data estimation. Due to time-varying characteristics of power line channels, the CIR needs to be constantly estimated. Xie and Li proposed a new precoded turbo equalization scheme for power-line communication channel [11]. The scheme transforms the non-recursive ISI channel to one that is recursive by precoding it with an appropriate rate recursive precoder, hence enabling the renowned interleaving gain.
A real-time Field-Programmable Gate Array (FPGA-based) architecture designed to implement channel estimation and equalization in broadband PLC has been proposed [12]. The approach is based on Orthogonal Frequency Division Multiplexing modulation (Wavelet-OFDM), also known as Filter Bank Multi-Carrier (FBMC), where cross-correlation techniques have been applied on the Zadoff-Chu sequences used as pilot signals or preambles. Many other studies regarding equalization of PLC have been presented in [17 -27].
From above paragraph, it has been noted that, only few researches concerning equalization in PLC have been done so far. Therefore more researches need to be conducted in this area in order to address properly the problem of equalization.
Therefore this paper presents adaptive decision feedback equalization techniques for broadband power line communication. Dissimilar power line channel models have been studied, however in this paper; the one presented by Anatory and others is used [13]. This model is suitable for a transmission line with multiple branches at a single node. The performance of the adaptive decision feedback equalizer is compared to blind and adaptive linear equalizers based on Based on Symbol Error Rate (SER) and Signal to Noise Ratio (SNR). finally, this paper is divided into four sections were section one gives introduction, section two presents channel models, section three presents simulation and analysis of the equalizer and last section gives conclusion.

POWER LINE CHANNEL MODELS 2.1 Characteristic of Power line channel
This section presents power-line channel model used in this paper. The use of power line for communication presents a number of technically difficult challenges. In addition to large attenuation, the power line cable network is one of the most electrically contaminated environments [14]. PLNs have been assembled with a variety of materials and cross sections are joined almost at random. This means that the inductive reactance along a wire itself will render a wide range of characteristic impedances at different points in the networks. Again, the PLN terminal impedance varies both at communication signal frequencies and with the time of day as the network's electrical load pattern varies [14]. All these contribute to stochastic channel changing behavior of the PLC channel. Hence the transfer characteristic (transfer function), () Hf of the channel is varied with respect to time.

Power-line Channel Models
The modeling of the PLC channel is of fundamental importance, since the quality of the transmission is highly influenced by the characteristics of the channel itself [15]. For investigating the performance of power line, a number of channel models for PLC have been proposed by researchers. However, in this paper a PLC channel model proposed by Anatory is used [13]. This model is suitable for a transmission line with multiple branches at a single node. The generalized channel transfer function is as represented as: where, N is the total number of branches connected at any node and terminated in any arbitrary load.
where S Z is the source characteristic impedance, Lr Z is characteristic load impedance, S V is the source voltage and () m Hf is the channel transfer function.The inductance, capacitance, resistance and conductance are calculated [13] as in (5), (6), (7) and (8): The parameter  is permeability in free space, D is the cable diameter, a is radius of conductor,  is permittivity in free space,  is the conductivity of material, and  is the depth factor. The characteristic impedance is calculated as: Since PLC uses overhead transmission lines and the separation between two lines is free space, the conductance is assumed to be zero [13]. The propagation constant is calculated as in (10), where  is the attenuation constant and  is the phase factor. With one branch PLC network the following parameters are used according to Anatory and others (Anatory et al, 2007). From Fig. 3

Fig. 4: Channel Impulse Response for PLC channel with two cascaded Branches
The Channel Impulse Response (CIR) parameters obtained from this section are used in simulation of to examine the performance of adaptive decision feedback equalizer.

Simulation and analysis of Adaptive Decision-Feedback Equalizer
This section presents the simulation of an Adaptive Decision feedback adaptive equalizer as shown in Fig. 5 and the PLC channel is applied through it [16]. Data symbols     , is generated. The error is then used to adjust the coefficients of the feedback and feedforward filters. Fig. 7 and Fig. 8 below shows simulation results for PLC channel with one branch and two networks respectively. In both cases, the received signal is corrupted with ISI, and then the received signal is equalized using ADFE. The results show that, with ADFE, the equalized sequence seemed to be equalized closer to the original signal. Thus, based on the scattered diagram it can be observed that, ADFE has removed considerable amount of ISI.   Fig. 10. Fig. 9 shows a plot of SER vs SNR for three equalizers when a PLC network with one branch is used. it can be observed from the simulation results that, the values of SER for the three equalizers decrease with increase in SNR (EO/NO) values. But the way the SER decreases is reasonably different from each other. The two equalizers (blind equalizer and adaptive linear equalizer (ALE)) shows poor result compared to ADFE. Therefore based on the result of Fig. 9, it shows that ADFE is the best equalization technique to be employed with PLC channels. Fig. 10 shows a plot of SER vs SNR for three equalizers when a PLC network with two cascaded branches is used. Comparing simulation results of the three equalizers given in Fig. 10, again it shows that, values of SER for all three equalizers decrease with increase in SNR (EO/NO) values. From Fig. 10, ADFE clearly shows better performances compared to ALE and blind equalizer although ALE seemed to give better result than blind equalizer. This proves that, even for SER, ADFE is still doing better compared to ALE and blind equalizer for equalization of PLC channel. Therefore this paper proposes ADFE as the best equalization technique for PLC channel.

Conclusion
In this paper, an Adaptive Decision Feedback Equalisation technique has been proposed to overcome the effect of stochastic channel changes in powerline communication systems. Appropriate power-line channel model was selected and channel impulse responses are obtained from the selected channel model. The impulse responses are used for simulating and analysing the performance of adaptive decision feedback equalizer. Based on Symbol Error Rate (SER) and Signal to Noise Ratio (SNR), the performance of adaptive decision feedback equalizer is compared to other two equalizers. Simulation results have proved that the adaptive equalization can perform better for equalization of power-line network compared to other techniques.