Pilot-Aided Channel Estimation on SC-PAM Based Visible Light Communications Omer Narmanlioglu 1 , Bugra Turan 2 , Refik Caglar Kizilirmak 3 , Sinem Coleri Ergen 2 , and Murat Uysal 1 1 Department of Electrical and Electronics Engineering, Ozyegin University, Istanbul, Turkey E-mail: omer.narmanlioglu@ozu.edu.tr, murat.uysal@ozyegin.edu.tr 2 Department of Electrical and Electronics Engineering, Koc University, Istanbul, Turkey E-mail: bturan14@ku.edu.tr, sergen@ku.edu.tr 3 Department of Electrical and Electronics Engineering, Nazarbayev University, Astana, Kazakhstan E-mail: refik.kizilirmak@nu.edu.kz Abstract—Estimation of the time-varying optical wireless chan- nel response is crucial in order to decode received signals coher- ently. In this work, we investigate symbol-error-rate and mean absolute error performance of different interpolation techniques including linear, nearest, spline, and piece-wise cubic Hermite interpolating polynomial (pchip), which are used in pilot-aided channel estimation process for visible light communication. The performance of interpolators is evaluated in realistic time-varying channel model, generated on Zemax and compared with each other under the consideration of different modulation orders, different pilot symbol periods, and different user equipment (UE) speeds through Monte Carlo simulations. The results reveal that spline and pchip techniques are more robust to low pilot symbol transmission rate and fast time-varying channel conditions as a consequence of high UE speeds. However, low complex linear interpolation technique can be chosen for highly rated pilot signal transmission cases or when optical channel varies slowly over the time. Index Terms—Channel Estimation, Interpolation, Visible Light Communication. I. I NTRODUCTION Visible light communication (VLC) is based on the dual usage of light emitting diodes (LEDs) for both illumination and communication purposes. Since LEDs can be modulated at high speeds that are not noticeable to human eye, they can be used as wireless access points [1]. VLC has been receiving attention as a short-range broadband wireless communication technology. This is likely to further grow along with the recent standardization efforts [2]. Fast time-varying optical wireless channel responses are known to lead error rate increments and error floors, resulting with system performance degradation. Hence, mitigating these degradation is key for reliable transmissions. In order to overcome fast-time varying effects of the channel, dynamic estimation of the optical wireless channel between LEDs and photodetectors (PDs) on the user equipment (UE) is necessary. Dynamic channel estimation aims to equalize the received signal effectively, before demodulation process. The channel The work of M. Uysal was supported by the Turkish Scientific and Research Council (TUBITAK) under Grant 215E311. Sinem Coleri Ergen acknowledges the financial support by the Turkish Academy of Sciences (TUBA) within the Young Scientist Award Program (GEBIP) and METU-Prof. Dr. Mustafa Parlar Foundation Research Encouragement Award. estimation can be performed by inserting pilot signals to the transmitted frame structure with a certain period that is known by the receiver side. Therefore, receiver first estimates the channel response with the certain period, then interpolates time-varying channel response using a specific interpolation technique. Common interpolation techniques such as linear, nearest, spline interpolation, or piece-wise cubic Hermite in- terpolating polynomial (pchip) have different implementation and computational complexities and offer different accuracy with respect to the interpolated data type. Various optical wireless channel estimation schemes in the context of VLC are investigated in the literature [3]– [12]. Authors in [3] proposed an adaptive equalization system for on-off keying (OOK) modulated multiple input multiple output (MIMO) VLC system. In [4]–[6], authors analyzed the performance of Least Square (LS) and minimum mean square error (MMSE) channel estimation techniques over direct cur- rent biased optical OFDM (DCO-OFDM) based single-input single-output (SISO) VLC scheme. In [7], authors proposed a discrete Fourier transform (DFT) based post-processing channel estimation to enhance the performance of conventional LS estimator. Similarly, [8] proposed an adaptive channel es- timation algorithm for asymmetrically-clipped optical OFDM (ACO-OFDM) based VLC. Moreover, [9] investigate the impact of different one-dimensional interpolation technique on pilot-aided channel estimation over ACO-OFDM based SISO VLC. [10] proposed sparse Bayesian relevance vector machine regression based channel estimation for orthogonal frequency-division multiplexing (OFDM) based VLC. In [11] and [12], a statistical Bayesian MMSE channel estimator is presented for SISO DCO-OFDM system. However, none of the above works consider realistic time-varying response of optical VLC channel and relatively they cannot perform any estimation process over time-varying channels. In this work, we investigate the symbol-error-rate (SER) and mean absolute error (MAE) performance of different interpolation techniques with respect to varying modulation orders, pilot signal periods, and UE speeds over the realistic time-varying channel model that has been proposed in our previous work [13]. The remainder of this work is organized as follows. In Section II, we describe the mobile channel model under con-