Yazar "Abut, Fatih" seçeneğine göre listele

Listeleniyor 1 - 6 / 6
  • [ X ]
    Öğe
    A distributed measurement architecture for inferring TCP round-trip times through passive measurements
    (Tubitak Scientific & Technological Research Council Turkey, 2019) Abut, Fatih
    The round-trip time (RTT), defined as the time elapsed for transmission of a data packet to travel from one endpoint to the other and back again, is an important parameter for Internet quality. This paper proposes an extended version of the well-known SYN/ACK (SA) methodology for passively measuring the RTTs over Transmission Control Protocol (TCP) connections. Differently from the original version of the SA methodology and the rest of studies in the related literature, the proposed passive methodology measures not only the total RTT of an end-to-end connection but also the proportion of the existing connection sections on this entire RTT in a passive way if the connection between client and server is established via intermediate stations. A distributed measurement architecture has been designed that implements the extended SA methodology. Through tests in a controlled laboratory environment, various verification and performance evaluation experiments were conducted to determine the accuracy level of the measurement technique and how the distributed architecture behaves regarding resource requirements as the amount of incoming network traffic increases. Accuracy verification experiments show that on average about 92.66% of the passive measurements are within 10% or 5 ms, whichever is larger, of the RTT that ping would actively measure. Furthermore, the results reveal that using today's commodity hardware, the designed distributed architecture exhibits acceptable satisfactory scaling performance and can practically be used to passively measure RTTs of each hop within medium-sized communication networks.
  • [ X ]
    Öğe
    A robust ensemble feature selector based on rank aggregation for developing new VO2max prediction models using support vector machines
    (Tubitak Scientific & Technological Research Council Turkey, 2019) Abut, Fatih; Akay, Mehmet Fatih; George, James
    This paper proposes a new ensemble feature selector, called the majority voting feature selector (MVFS), for developing new maximal oxygen uptake (VO(2)max) prediction models using a support vector machine (SVM). The approach is based on rank aggregation, which meaningfully utilizes the correlation among the relevance ranks of predictor variables given by three state-of-the-art feature selectors: Relief-F, minimum redundancy maximum relevance (mRMR), and maximum likelihood feature selection (MLFS). By applying the SVM combined with MVFS on a self-created dataset containing maximal and submaximal exercise data from 185 college students, several new hybrid VO(2)max prediction models have been created. To compare the performance of the proposed ensemble approach on prediction of VO(2)max, SVM-based models with individual combinations of Relief-F, mRMR, and MLFS as well as with other alternative ensemble feature selectors from the literature have also been developed. The results reveal that MVFS outperforms other individual and ensemble feature selectors and yields up to 8.76% increment and 11.15% decrement rates in multiple correlation coefficients (Rs) and root mean square errors (RMSEs), respectively. Furthermore, in addition to reconfirming the relevance of sex, age, and maximal heart rate in predicting VO(2)max, which were previously reported in the literature, it is revealed that submaximal heart rates and exercise times at 1.5-mile distance are two further discriminative predictors of VO(2)max. The results have also been compared to those obtained by a general regression neural network and single decision tree combined with MVFS, and it is shown that the SVM exhibits much better performance than other methods for prediction of VO(2)max.
  • [ X ]
    Öğe
    Comparison of CoAP and CoCoA Congestion Control Mechanisms in Grid Network Topologies
    (2018) Demir, Alper Kamil; Abut, Fatih
    The Internet of Things (IoT) is a vision of the future Internet. Due to limited resources of IoT devices, a new generationof protocols and algorithms are being developed and standardized. The Constrained Application Protocol (CoAP) hasbeen designed by the Internet Engineering Task Force (IETF) for application layer communication. CoAP is based onUser Datagram Protocol (UDP), a simple transport layer protocol that does not handle congestion within the network.However, the phenomenon of congestion in IoT networks is also a major problem. Thus, the core CoAP specificationoffers a basic CoAP congestion control (CC) mechanism based on retransmission timeout (RTO) with binaryexponential backoff (BEB). Default CoAP CC is insensitive to network conditions. Thus, to improve the default CoAPCC, CoAP Simple Congestion Control/Advanced (CoCoA), defined in a draft specification, is being standardized by theIETF CoRE working group. Nevertheless, comparison of default CoAP CC and CoCoA has not been sufficientlyinvestigated in the literature. In this paper, we investigate and present comparison of default CoAP CC and CoCoA interms of throughput (i.e. number of requests/second) by varying number of concurrent clients where each clientcontinuously sends back-to-back traffic to servers residing in 1x6, 3x6 and 5x6 grid network topology. Our results showthat CoCoA is not always better than default CoAP CC in terms of throughput in some scenarios. As a result, designand development of new CoAP CC mechanisms are open to research.
  • [ X ]
    Öğe
    Data-driven modelling and prediction of coap throughput in a grid network topology
    (TUBITAK, 2020) Demir, Alper Kamil; Abut, Fatih
    In this study, we propose new models for predicting the average throughput in a 4x4 grid Constrained Application Protocol (CoAP)-based IoT network using Support Vector Machine (SVM) and Multiple Linear Regression (MLR). Two different CoAP congestion control mechanisms have been considered: the default CoAP congestion control (CC) and the CoAP Simple Congestion Control/Advanced (CoCoA). On the client-side, we run 3, 6, 9, 12 or 15 CoAP clients requesting packets, sized with 12, 24, 36 or 48 bytes, from different CoAP servers over 4x4 grid IoT network configured with packet delivery ratios of 90, 95 or 100. In total, 60 different experimental scenarios, each of which was run 10 times to determine the average throughput of default CoAP CC and CoCoA clients, were created. Using 10-fold cross-validation, the performance of the prediction models has been evaluated using several performance metrics. The results show that combining packet delivery ratio and number of concurrently sending clients in a model leads to the highest correlation with the average CoAP throughput of the IoT network. Particularly, this model produces the lowest prediction error among all SVM-based and MLR-based models, regardless of whether the default CoAP CC or CoCoA is used as the congestion control mechanism. © 2020, TUBITAK. All rights reserved.
  • [ X ]
    Öğe
    mlCoCoA: a machine learning-based congestion control for CoAP
    (Tubitak Scientific & Technological Research Council Turkey, 2020) Demir, Alper Kamil; Abut, Fatih
    Internet of Things (IoT) is a technological invention that has the potential to impact on how we live and how we work by connecting any device to the Internet. Consequently, a vast amount of novel applications will enhance our lives. Internet Engineering Task Force (IETF) standardized the Constrained Application Protocol (CoAP) to accommodate the application layer and network congestion needs of such IoT networks. CoAP is designed to be very simple where it employs a genuine congestion control (CC) mechanism, named as default CoAP CC leveraging basic binary exponential backoff. Yet efficient, default CoAP CC does not always utilize the network dynamics the best. As a result, CoCoA has been exposed to better utilize the IoT networks. Although CoCoA considers the network dynamics, the RTO calculation of CoCoA is based on constant coefficient values. However, our experiments show that these constant values, in general, do not achieve the best throughput. Inspired by these observations, we propose a new machine learning-based CC mechanism called as mlCoCoA that is a variation of CoCoA. Particularly, mlCoCoA sets retransmission timeout (RTO) estimation parameters of CoCoA adaptively by using a machine learning method. In this study, we applied support vector machines on a self-created dataset to develop new models for improving the throughput of the IoT network with dynamic selection of CoCoA coefficient values. We carried out extensive simulations in Cooja environment coupled with Californium. Our results indicate that compared to the performance of default CoAP CC and CoCoA mechanisms, mlCoCoA has merit in terms of improving the throughput of CoAP applications.
  • [ X ]
    Öğe
    User-level Performance Evaluation of VoIP under Different Background TCP Traffic Conditions in ns-2
    (2019) Abut, Fatih
    Voice over IP (VoIP) is gaining more and more importance and displaces the traditional telephony. For example, more than 300 millionmonthly active users worldwide use the popular VoIP application \"Skype\". However, a big problem in the VoIP environment is the voicequality. The purpose of this study is to investigate the effects of background TCP traffic on perceived voice quality of a VoIP conversationat the user-level using the G.711 codec. Two different playout buffering policies including static buffering and optimal buffering havebeen applied by the VoIP server. For comparison purposes, the same experiments have also been repeated when no playout bufferingpolicy has been used by the VoIP server. A three-hop network topology consisting of a source, a transit, and a destination subnetworkwas simulated whereas the end-to-end capacity of the entire network was limited by a 1.5 Mbps Asymmetric Digital Subscriber Line(ADSL) link. Multiple simultaneous TCP connections with different segment sizes were established to simulate the various conditionsof background traffic. By using the ns2voip framework, an enhancement to the popular Network Simulator 2 (ns-2), extensive simulationexperiments for analyzing the VoIP user-level performance have been carried out and the voice quality has been evaluated by calculatingthe Mean Opinion Score (MOS). The results show that the voice quality is strongly negatively affected by background TCP traffic, evenin the presence of a single TCP flow with 1500 Byte segments. Also, the size of background TCP segments significantly influences theachievable MOSs of VoIP conversations. However, it has also been observed that aggregating multiple speech frames into a single IPpacket can increase the MOS. Particularly, depending on the number and segment size of background TCP flows, aggregation of theoptimal number of speech frames into the same IP packet improves the MOSs up to 14.61% over a 1.5 Mbps ADSL link.