Browsing by Author "Kabakulak, Banu."

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Design and analysis of communication systems with high error correction capability through optimization
    (Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2018., 2018.) Kabakulak, Banu.; Taşkın, Zeki Caner.
    Channel coding is the term used for the collection of techniques that are employed in order to minimize errors which occur during the transmission of digital information from one place to another. Low–density parity–check (LDPC) code family takes at tention with its channel capacity–approaching error correction capability and sparse parity–check matrix representation. Sparsity property of the matrix gives rise to the development of heuristic iterative decoding algorithms with low complexity. Ease of the application of iterative decoding algorithms brings the advantage of low decod ing latency. In spite of these benefits of LDPC codes, receiver can obtain erroneous information because of both structural properties of LDPC codes and non–optimal decoders. In the first part of this thesis, we develop optimization–based LDPC decoding algorithms for a communication system with high error performance and we compare its performance with the existing methods in the literature. Error performance of a communication system can still be improved by determining and eliminating small cycles in LDPC codes that cause iterative decoding algorithms to halt or terminate without a conclusive result during the decoding process. At the second place, we implement heuristic and optimization–based approaches for efficiently designing high quality LDPC codes of practically relevant dimensions. We carry out extensive com putational experiments to assess the efficiency of proposed methods.
  • Thumbnail Image
    Item
    Optimal placement, scheduling and routing to maximize lifetime in wireless sensor networks under connectivity restrictions
    (Thesis (M.S.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2010., 2010.) Kabakulak, Banu.; Altınel, İ. Kuban.
    A wireless sensor network consists of distributed autonomous electronic devices called sensors. They are capable of sensing the changes in their vicinity, process the information as data packets and transmit the data to other sensors or a base station namely sink. In order to have an effective sensor network that can keep track of the changes in the interested region, sensors have to work cooperatively since they have limited battery energy. Working in accordance is also important to transmit the collected information eventually to a sink, since sensors can communicate only with the others that fall in a certain range. In most of the real life applications, for a wireless sensor network the number of periods that the network can operate as desired is a significant performance indicator. In this thesis, we propose mixed-integer linear programming models to maximize the network lifetime by optimally determining the locations of sensors, activity schedules of the deployed sensors, sink assignments of the active sensors and their data flow routes to the corresponding sink over a finite planning horizon subject to coverage, flow conservation, energy consumption and budget constraints. Then, we introduce valid inequalities to solve the problem easily. Due to the characteristics of the problem, even the small instances cannot be solved exactly in considerable amount of time and the linear programming relaxations give poor upper bounds. Hence, we develop heuristics using techniques such as Lagrangean relaxation and greedy selection criterion. Computational experiments indicate that the heuristic methods are accurate and efficient.