Abstract-In this paper, we experimentally investigate the statistical distribution of intensity fluctuations for underwater wireless optical channels under different channel conditions, namely fresh and salty underwater channels with and without air bubbles. To do so, we first measure the received optical signal with a large number of samples. Based on the normalized acquired data the channel coherence time and the fluctuations probability density function (PDF) are obtained for different channel scenarios. Our experimental results show that salt attenuates the received signal while air bubbles mainly introduce severe intensity fluctuations. Moreover, we observe that log-normal distribution precisely fits the acquired data PDF for scintillation index (σ 2 I ) values less than 0.1, while Gamma-Gamma and K distributions aptly predict the intensity fluctuations for σ 2 I > 1. Since neither of these distributions are capable of predicting the received irradiance for 0.1 < σ 2 I < 1, we propose a combination of an exponential and a log-normal distributions to perfectly describe the acquired data PDF for such regimes of scintillation index.

No abstract

No abstract

In this paper, we present an optical computing method for string data alignment applicable to genome information analysis. By applying moiré technique to spatial encoding patterns of deoxyribonucleic acid (DNA) sequences, association information of the genome and the expressed phenotypes could more effectively be extracted. Such moiré fringes reveal occurrence of matching, deletion and insertion between DNA sequences providing useful visualized information for prediction of gene function and classification of species. Furthermore, by applying a cylindrical lens, a new technique is proposed to map twodimensional (2D) association information to a one-dimensional (1D) column of pixels, where each pixel in the column is representative of superposition of all bright and dark pixels in the corresponding row. By such a time-consuming preprocessing, local similarities between two intended patterns can readily be found by just using a 1D array of photodetectors and postprocessing could be performed on specified parts in the initial 2D pattern. We also evaluate our proposed circular encoding adapted for poor data alignment condition. Our simulation results together with experimental implementation verify the effectiveness of our dynamic proposed methods which significantly improve system parameters such as processing gain and signal to noise ratio (SNR).

Recent advances in wearable devices and Internetof-Things (IoT) have led to massive growth in sensor data generated in edge devices. Labeling such massive data for classification tasks has proven to be challenging. In addition, data generated by different users bear various personal attributes and edge heterogeneity, rendering it impractical to develop a global model that adapts well to all users. Concerns over data privacy and communication costs also prohibit centralized data accumulation and training. This paper proposes a novel personalized semi-supervised federated learning (SemiPFL) framework to support edge users having no label or limited labeled datasets and a sizable amount of unlabeled data that is insufficient to train a well-performing model. In this work, edge users collaborate to train a hyper-network in the server, generating personalized autoencoders for each user. After receiving updates from edge users, the server produces a set of base models for each user, which the users locally aggregate them using their own labeled dataset. We comprehensively evaluate our proposed framework on various public datasets and demonstrate that SemiPFL outperforms state-of-art federated learning frameworks under the same assumptions. We also show that the solution performs well for users without labeled datasets or having limited labeled datasets and increasing performance for increased labeled data and number of users, signifying the effectiveness of SemiPFL for handling edge heterogeneity and limited annotation. By leveraging personalized semi-supervised learning, SemiPFL dramatically reduces the need for annotating data and preserving privacy in a wide range of application scenarios, from wearable health to IoT.