MISO visible light communication system utilizing hybrid post-equalizer aided pre-convergence of STBC decoding Download: 626次
Over the past few years, visible light communication (VLC) has received significant attention to moving parts of high-speed indoor data transmission. Unlike radio frequency (RF) wireless communication, VLC utilizes the already installed light emitting diode (LED) as data transmitters (Tx)[1
However, limited modulation bandwidth is one of the obstacles to improve the transmission rate for the VLC system[7–
In the multi-dimensional system, linear damage will lead to the intersymbol interference (ISI)[16,17]. Some linear post-equalization algorithms, such as recursive least square (RLS) and least mean square (LMS) act as the savior to eliminate ISI. However, these two equalization algorithms require a certain length of training sequence (TS). Instead, some researchers proposed blind equalization to adaptively revise distorted signals, such as multi-modulus algorithm (MMA), cascaded-MMA (CMMA) and the modified CMMA (MCMMA)[18].
The LED nonlinearity by electrical amplifiers (EAs) and optoelectronic devices also distorts signals and deteriorates the system performance. Therefore, the compensation for nonlinearity has gradually become one of the significant challenges in the VLC system. Pre-distortion and post-distortion techniques generally go hand-in-hand, and both are valid solutions to mitigate the VLC channel artifacts[19,20]. However, the above two literatures are only validated by simulation results and lack of experimental demonstration. As a promising solution, the Volterra-series-based equalizer plays a role of significance to mitigate the LED nonlinearity[21]. In our previous work, a blind post-equalization scheme called cascaded Volterra modified MMA was employed and demonstrated to compensate for linearity and mitigate the LED nonlinearity in a carrierless amplitude phase (CAP) modulation-based VLC system[22]. The authors in Ref. [23] employed MCMMA to equalize received signals in the MISO-STBC system. The experiments demonstrate that it also aids pre-convergence of STBC decoding at the receiver. However, the compensation for nonlinearity is never considered.
In this Letter, we present a modified hybrid linear and nonlinear post-equalizer to mitigate the nonlinear distortion based on our previous work[23]. Extensive lab experiments demonstrate that the hybrid post-equalizer is also an excellent complementary technology for STBC decoding. Meanwhile, we propose an optimization algorithm for channel estimation, as the existing method in Ref. [23] is suboptimal. With these improved algorithms, a data rate of
First, we formulate an MISO-STBC model with the modulation of QAM to investigate the proposed hybrid post-equalizer performance. An indoor VLC-MISO system has the number of LEDs and PDs of
Before the in-phase and quadrature (I–Q) modulation, Alamouti’s STBC schemes encode the data as follows:
The VLC-MISO model can be expressed by
In this model, two QAM symbols encoded by STBC are mapped into sixteen constellations. As the symbols must be non-negative and real-valued, the received signals of 49-QAM are generated, as it employed instant messaging (IM) via LEDs for transmission and DD via PDs for recovery. However, the forty-nine constellations are rolled into a mass by the VLC-LOS propagation loss, and then, the signals cannot be recovered by STBC decoding.
The MMA is one of the classic and simple blind equalizations to avoid extra phase rotation[24]. In parallel, the nonlinearity is more obvious as to higher modulation orders appearing, which is also a Gordian knot in the multi-dimensional VLC system with few investigated before. The Volterra series theory has been verified as an alternative way to compensate the nonlinearity. In this article, a modified hybrid Volterra-MCMMA based on Ref. [22] is proposed.
Theoretically, if more terms are used to estimate the original symbols, the results will be more accurate. In fact, the complexity of the Volterra series is too high. As a tradeoff between equalization performance and computational complexity, only the second order terms are taken into consideration[25]:
The hybrid equalizer includes linear equalization and nonlinear equalization:
Wherein, the output of
Unlike the method in Ref. [22], the coefficients of the transfer function of I–Q components will be updated individually. Its cost function is expressed as
At the receiver, the modified hybrid post-equalizer block diagram can be structured, as in Fig.
In this model, we need the channel matrix to decode received symbols. The recovery signals can be expressed as
To access to slightly accurate channel information, we insert a shorter length of TSs in front of STBC symbols.
We get a better channel matrix according to vast calculations offline and the analysis:
Wherein, the magnitude of
In principle, the longer the TSs length, the easier for STBC decoding. Actually, as we concluded recently, the length of the TS had little effect on the
Subsequently, we assume
Wherein,
As analyzed formerly, the loss in the VLC-LOS channel can be assumed to be the same. However, different elements of the matrix experience different losses. The purpose of Eq. (
Finally, we employ
Figure
First, we investigate the BER versus signal strength of different LEDs without the proposed channel estimation algorithm.
As is shown in Fig.
Fig. 3. (a) BER versus DC voltages of LED1; (b) BER versus DC voltages of LED2; (c) BER versus VPP voltages of LED1; (d) BER versus VPP voltages of LED2.
The following two figures depict the BER versus the signal drive voltages (peak-to-peak voltage, VPP) of one slide in the range of 0.31 to 0.4 V, while the other is fixed at 0.35 V. These two figures show that it permits an interval of 0.06 V without resisting the nonlinearity. However, an allowable interval is beyond 0.09 V with the hybrid post-equalizer. We also find that the BER performance is the best with the same voltage.
Subsequently, we investigate the performance of the proposed channel estimation algorithm, which is shown in Fig.
Fig. 4. (a)–(c) 49 constellations before STBC decoding of (1), (2), and (3); (a-1)–(c-1) 16 constellations after STBC decoding without channel optimization of (1), (2), and (3); (a-2)–(c-2) 16 constellations after STBC decoding with channel optimization of (1), (2), and (3).
For the sake of clarity, the received constellations of different equalizers are: (1) without any post-equalizer; (2) only with MCMMA; (3) with the proposed hybrid post-equalizer. Wherein, the signals before STBC decoding are shown in Figs.
From former works, we conclude that the signals cannot be decoded without any equalization. In the following works, we utilize the case of a pure linear equalizer without proposing an optimization algorithm as the object being compared.
The baud rate in different cases is compared in Fig.
For clear comparison, we bring the concept of the Q factor, which is inversely proportional to the BER. We measure the Q factor versus the baud rate with the aforementioned methods. Figure
Based on the aforementioned works, we draw the following conclusions: the case of combining with the modified hybrid equalizers and proposed channel estimation algorithm uniformly outperform the case of the pure linear equalizer. However, the signals cannot be decoded by the proposed channel matrix with STBC decoding in the case of different signal strengths in both LEDs.
In this Letter, we first present a modified hybrid post-equalizer against both linear distortion and nonlinear distortion. Meanwhile, we design a better channel estimation algorithm to improve the accuracy of STBC decoding. The MISO-STBC model is formulated to verify the correctness of the proposed two algorithms. A transmitting rate of
[1]
[2]
[3]
[4]
[5]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
Liang Qiao, Xingyu Lu, Shangyu Liang, Jiao Zhang, Nan Chi. MISO visible light communication system utilizing hybrid post-equalizer aided pre-convergence of STBC decoding[J]. Chinese Optics Letters, 2018, 16(6): 060604.