Source author record

Bugra Turan

Bugra Turan appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

cond-mat.mtrl-sci eess.SP Machine Learning Networking and Internet Architecture

Catalog footprint

What is connected

4works

4topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2021arXiv

Vehicular Visible Light Communications for Automated Valet Parking

Visible light communication (VLC) is a promising Optical Wireless Communications (OWC) scheme that is demonstrated to provide secure, line-of-sight (LoS), and short-distance vehicle-to-vehicle (V2V) and vehicle-to-infrastructure(V2I) communications. Recently, automated driving applications, supported by V2I links are proposed to increase the reliability of the autonomous vehicles. To this regard, we propose a VLCbased V2I scheme to increase the V2I communication redundancy of autonomous valet parking (AVP) applications, through jam-free and location-based characteristics of VLC. In this paper, we demonstrate a novel architecture to support indoor parking-garage online-map update with vehicle on-board data transmissions and location-based map update dissemination through bidirectional VLC communications. The proposed system yields error-free LoS transmissions with Direct Current Biased Optical OFDM (DCO-OFDM) up to 33 m transmitter-receiver distance enabling vehicle CAN Bus data, infrastructure camera video, and LIDAR point cloud data sharing in an indoor parking garage.

preprint2020arXiv

Machine Learning Based Channel Modeling for Vehicular Visible Light Communication

Optical Wireless Communication (OWC) propagation channel characterization plays a key role on the design and performance analysis of Vehicular Visible Light Communication (VVLC) systems. Current OWC channel models based on deterministic and stochastic methods, fail to address mobility induced ambient light, optical turbulence and road reflection effects on channel characterization. Therefore, alternative machine learning (ML) based schemes, considering ambient light, optical turbulence, road reflection effects in addition to intervehicular distance and geometry, are proposed to obtain accurate VVLC channel loss and channel frequency response (CFR). This work demonstrates synthesis of ML based VVLC channel model frameworks through multi layer perceptron feed-forward neural network (MLP), radial basis function neural network (RBF-NN) and Random Forest ensemble learning algorithms. Predictor and response variables, collected through practical road measurements, are employed to train and validate proposed models for various conditions. Additionally, the importance of different predictor variables on channel loss and CFR is assessed, normalized importance of features for measured VVLC channel is introduced. We show that RBF-NN, Random Forest and MLP based models yield more accurate channel loss estimations with 3.53 dB, 3.81 dB, 3.95 dB root mean square error (RMSE), respectively, when compared to fitting curve based VVLC channel model with 7 dB RMSE. Moreover, RBF-NN and MLP models are demonstrated to predict VVLC CFR with respect to distance, ambient light and receiver inclination angle predictor variables with 3.78 dB and 3.60 dB RMSE respectively.

preprint2016arXiv

From leaf to tree: upscaling of artificial photosynthesis

Energy storage becomes crucial for energy systems with an increasing share of renewable energy sources. Artificial photosynthesis, in particular photovoltaic water splitting, provides both sustainable energy generation and energy storage in the form of hydrogen. However, only a few concepts for scalable devices were reported in the literature. Here, we introduce a new concept which, by design, is scalable and compatible with every thin-film photovoltaic technology. The concept allows for independent geometrical optimization of the photovoltaic and the electrochemical part. The scalability is achieved by continuous mirroring of a base unit. We demonstrate a fully integrated, wireless device with a stable and bias-free operation for 40 hours. The concept was scaled to an area of 64 cm2 comprising 13 base units and exhibited a solar-to-hydrogen efficiency of 3.9%. The concept and its successful realization is an important contribution towards the large scale application of artificial photosynthesis.

preprint2015arXiv

A concept for Lithography-free patterning of silicon heterojunction back-contacted solar cells by laser processing

Silicon heterojunction (SHJ) solar cells with an interdigitated back-contact (IBC) exhibit high conversion efficiencies of up to 25.6%. However, due to the sophisticated back-side pattern of the doped layers and electrode structure many processing and patterning steps are required. A simplification of the patterning steps could ideally increase the yield and/or lower the production costs. We propose a patterning approach for IBC SHJ solar cells free of any photo-lithography with the help of laser-induced forward transfer (LIFT) of the individual layer stacks to create the required back-contact pattern. The concept has the potential to lower the number of processing steps significantly while at the same time giving a large degree of freedom in the processing conditions optimization of emitter and BSF since deposition of the intrinsic/doped layers and processing of the wafer are all independent from each other.