Melih Gunay

CAR Approach for the Internet of Things

In this paper, we propose a novel context-aware routing (CAR) approach that uses the cloud as an extra level of data-request processing to improve the network performance in terms of data delivery. Data delivery in the Internet of Things depends heavily on numerous factors, such as the amount of data, end-to-end in-network delay, and setup time. The CAR approach is significantly improving the current request-response model, especially while the exchanged in-network data amount increases and data are sent from source to destination in a peer-to-peer fashion. What we are trying to show in this paper, in particular, is the benefits of having a central context-aware server (in the cloud) in improving the end-user experience. Hence, the proposed CAR approach is a typical candidate for data-intensive cloud-based applications. It considers source and destination requirements in terms of data size, delay, link capacity, and available applications on the operating devices as well. Extensive simulations are performed, and achieved results show the efficiency of our approach against other competitive approaches in terms of in-network delay and packet delivery ratio.

Path planning for mobile DCs in future cities

In future smart-cities, public transportation vehicles are planned to serve as data couriers (DCs) in order to exchange massive amounts of data chunks. In this research, we study the path planning problem for these DCs while optimizing their counts and their total traveled distances. As the total collected load on a given DC route cannot exceed its storage capacity, it is important to decide on the size of the exchanged data-packets (images, videos, etc.) and the sequence of the targeted data sources to be visited. We propose a hybrid heuristic approach for public data delivery in smart-city settings. In this approach, public vehicles are utilized as DCs that read/collect data from numerously distributed Access Points (APs) and relay it back to a central processing base-station in the city. We also introduce a cost-based fitness function for DCs election in the smart-city paradigm. Our cost-based function considers resource limitations in terms of DCs count, storage capacity, and energy consumption. Extensive simulations are performed, and the results confirm the effectiveness of the proposed approach in comparison to other heuristic approaches with respect to total traveled distances and overall time complexity.

The road to dynamic Future Internet via content characterization

The Internet evolved from a network with a few terminals to an intractable network of millions of nodes. Recent interest in information-centric networks (ICNs) is gaining significant momentum as a Future Internet paradigm. The key question is, hence, how to model the massive amount of connected nodes with their content requests in dynamic paradigm. In this paper, we present a novel method to characterize data requests based on content demand ellipse (CDE), focusing on efficient content access and distribution as opposed to mere communication between data consumers and publishers. We employ an approach of a promising eminence, where requests are characterized by type and popularity. Significant case studies are used to demonstrate that critical properties of ellipses may be used to characterize the content request irregularity during peak times. Depending on the degree of irregularity, the curve we plot becomes elliptic with a positive eccentricity less than one and an orientation centered with a bias. Real traffic data have been used to demonstrate how various data demand/request types affect eccentricity, orientation, and bias. Through simulations, we propose a dynamic resource allocation framework for Virtual Data Repeaters (VDRs) by correlating the resource allocation schema with the factors that affect the CDE in ICN.

Review on Machine Learning Based Lesion Segmentation Methods from Brain MR Images

Brain lesions are life threatening diseases. Traditional diagnosis of brain lesions is performed visually by neuro-radiologists. Nowadays, advanced technologies and the progress in magnetic resonance imaging provide computer aided diagnosis using automated methods that can detect and segment abnormal regions from different medical images. Among several techniques, machine learning based methods are flexible and efficient. Therefore, in this paper, we present a review on techniques applied for detection and segmentation of brain lesions from magnetic resonance images with supervised and unsupervised machine learning techniques.

Routing mobile data couriers in smart-cities

In this paper, we propose a new architecture to read the smart meters which are commonly distributed nowadays in smart cities. In this architecture, public transportation vehicles are utilized as Data Collectors (DCs) that reads these smart meters. Moreover, we target the path planning problem for these DCs given that a limited number of vehicles with a specific storage capacity are able to participate in collecting readings from these meters. We optimize the number of DCs while maintaining their minimum travelling distances and satisfied traffic constraints. We propose a Genetic-based Routing (GR) approach for more optimized solutions. Extensive simulation results are performed to confirm the effectiveness of the proposed approach in comparison to other heuristic approaches.

Adaptive Bias Field Correction: Application on Abdominal MR Images

Segmentation of medical images is one of the most important phases for disease diagnosis. Accuracy, robustness and stability of the results obtained by image segmentation is a major concern. Many segmentation methods rely on absolute values of intensity level, which are affected by a bias term due to in-homogeneous field in magnetic resonance images. The main objective of this paper is two folded: (1) To show efficiency of an energy minimization based approach, which uses intrinsic component optimization, on abdominal magnetic resonance images. (2) To propose an adaptive method to stop the optimization automatically. The proposed method can control the value of the energy functional and stops the iteration efficiently. Comparisons with two previous state-of-the art methods indicate a better performance of the proposed method.

Automated Detection of Facial Disorders (ADFD): a novel approach based-on digital photographs

Abnormal regions on a face are often indicative of facial dermatological diseases, such as rosacea, eczema, allergy, burn injury, facial rash and acne. Identification and quantitative evaluation of these facial disorders is often subjective and currently done by expert dermatologists. However, with the advances in image processing techniques, it is now possible to consistently identify, classify and objectively quantify facial skin disorders using digital photographs. Furthermore, digital photographs can be used to assess progression of the diseases by applying time-series analysis on infected (abnormal) facial regions of the photograph. In this paper, we propose a novel real-time approach for detection and segmentation of abnormal facial regions. Experimental results showed that the proposed method is efficient for segmentation of abnormal regions, which were caused by facial fever or disease, from digital photographs in terms of accuracy and required processing time.

TaqMan array card data management system for epidemiologic surveillance and clinical study

In this research, we have developed a web based TaqMan Array Cards (TMDK) Data Management System that responds to the needs of collecting, sharing and uploading data of many surveillance projects, and a Ct-value prediction algorithm to determine target gene existence in sample. Any epidemic case can rapidly be detected and comparative evaluations can be done for the results of different sites while maintaining data integrity since data access is provided through single point. Our Ct-value prediction algorithm resets the errors caused by individual based value identification while keeping Ct consistency. In our tests, with parallel calculations, we have observed that the critical operations can be completed under 30sec and analyzes could rapidly be done.

Priority based vehicle routing for agile blood transportation between donor/client sites

In this paper, we study Vehicle Routing Problem (VRP) for Blood Transporters (BTs) and propose an efficient vehicle routing scheme for blood transportation between hospitals or Donor/Client Sites (DCSs) within a region that is based on Artificial Intelligence. It is assumed that each BT in a fleet of vehicles starts and completes its route at a blood-bank while visiting a subset of DCSs using the shortest path. However, unlike traditional logistic planning, blood transportation may be time critical. Therefore, in our approach, the vehicle routing is formulated to take into account the urgency of the requests and responses. Consequently, the objective of this study is to minimize the number of BTs while maintaining their minimum traveling lengths considering priority. In this regards, we extended the classical Capacitated VRP (CVRP) and reformulated requests to take into account the priority by assigning weight to each request. A hybrid meta-heuristic algorithm including Genetic Algorithms and Local Search is used to simulate transporting blood requests of DCSs. We challenged our approach with symmetrical CVRP instances taken from literature. In this case study, we observed that both the cost and response time are reduced dramatically for emergency.

Machine Learning for Optimum CT-Prediction for qPCR

Introduction of fluorescence-based Real-Time PCR (RT-PCR) is increasingly used to detect multiple pathogens simultaneously and rapidly by gene expression analysis of PCR amplification data. PCR data is analyzed often by setting an arbitrary threshold that intersect the signal curve in its exponential phase if it exists. The point at which the curve crosses the threshold is called Threshold Cycle (CT) for positive samples. On the other, when such cross of threshold does not occur, the sample is identified as negative. This simple and arbitrary however not an elagant definition of CT value sometimes leads to conclusions that are either false positive or negative. Therefore, the purpose of this paper is to present a stable and consistent alternative approach that is based on machine learning for the definition and determination of CT values.