Kongu Engineering College

About: Kongu Engineering College is a based out in . It is known for research contribution in the topics: Cluster analysis & Control theory. The organization has 2001 authors who have published 1978 publications receiving 16923 citations.

Isolated Offline Tamil Handwritten Character Recognition Using Deep Convolutional Neural Network

Abstract: In this paper, a Convolutional Neural Network architecture (ConvNet) for offline isolated Tamil character recognition is proposed. A first-ever attempt has been made to recognize all 247 characters in the Tamil text using 124 unique symbols. The proposed architecture contains two Convolutional layers and Two Fully Connected (FC) Layers with ReLu activation function. Softmax function is used in the final layer to compute the probability of the classes. The 9.6 million parameters of the network are randomly initialized using He initialization and fine-tuned using Nesterov Accelerated batch gradient descent optimization algorithm. Dropout regularization method has been used to avoid over-fitting of the network to the training data. A total of 98,992 image samples from IWFHR database are divided into 69% for training set (68,488), 20% for validation set (20,584) and 11% for test set (9920). Cross entropy loss has been used during the training phase to measure the loss and thereby update the parameters of the network. The network has achieved 88.2% training accuracy and 71.1% testing accuracy. The reason for reduction in the test accuracy is analysed. The source code and the dataset have been published for a quicker reproducibility of the result.

Efficient anomaly detection and mitigation in software defined networking environment

Abstract: A Computer network or data communication network is a telecommunication network that allows computers to exchange data. Computer networks are typically built from a large number of network devices such as routers, switches and numerous types of middle boxes with many complex protocols implemented on them. They need to accomplish very complex tasks with access to very limited tools. As a result, network management and performance tuning is quite challenging. Software-Defined Networking (SDN) is an emerging architecture purporting to be adaptable, cost-effective, dynamic and manageable pursuing to be suitable for the high-bandwidth, changing nature of today's applications. SDN architectures decouples network control and forwarding functions, making network control to become directly programmable and the underlying infrastructure to be abstracted from applications and network services. The network security is a prominent feature of the network ensuring accountability, confidentiality, integrity, and protection against many external and internal threats. An Intrusion Detection System (IDS) is a type of security software designed to automatically alert administrators when someone or something is trying to compromise information system through malicious activities or through security policy violations. Security violation in SDN environment needs to be identified to prevent the system from an attack. The proposed work aims to detect the attacks on SDN environment. Detecting anomalies on SDN environment will be more manageable and efficient.

Achieving machining effectiveness for AISI 1015 structural steel through coated inserts and grey-fuzzy coupled Taguchi optimization approach

Abstract: Multi-objective optimization technique has become an essential step in the selection of cutting parameters. The intension of this research study is to analyze the performance characteristics of coated carbide inserts on their measured output responses during machining AISI 1015 steel. This paper targets to optimize the machining parameters such as speed, cutting depth, feed rate, cutting fluid flow rate, and coating material when multiple responses like surface roughness and flank wear were considered at the same time during turning. This research study also intends to examine scientifically the effect of machining parameters on quality measures during machining structural AISI 1015 steel. Cathodic arc evaporation–coated titanium aluminum nitride (TiAlN), titanium aluminum nitride/tungsten carbide-carbon (TiAlN/WC-C), and uncoated CNC inserts were used for the study. SEM and energy-dispersive X-ray analysis were performed to confirm the presence of coated elements. Micro-hardness was measured for coated, pure inserts, and TiAlN/WC-C-coated tool exhibited a higher hardness of 22.11 GPa compared with pure and coated tools. Five process parameters were used for this study, each at three stages. The experimental design was laid based on Taguchi’s L27 orthogonal array. In this research study, a multi-objective hybrid optimization technique comprising grey relation and fuzzy logic conjugated with the Taguchi design of experiments was used. The process parameters were optimized by grey relation analysis followed by fuzzification using Mamdani fuzzy engine and then optimized through Taguchi analysis. The parameter combination of speed 500 rpm, depth of cut of 1 mm, a feed rate of 0.05 mm/rev, cutting fluid flow rate at high level, and TiAlN/WC-C coating was found to be the optimum input parameters. The confirmatory test was also performed to validate the hybrid optimization approach.

Filtering spoofed traffic at source end for defending against DoS / DDoS attacks

Abstract: Tackling the challenge of distinguishing legitimate traffic from attack would aid in the detection of denial of service (DoS) / distributed DoS (DDoS) attacks. Spoofing of source address would further harden the detection of such attacks. In this paper, we propose a flow based scheme to detect the DoS attacks that adapts itself to the changes trends of the current traffic. The proposed system weeds out most of the spurious traffic at the source end, thus avoiding clogging of the target and the network. The proposed scheme distinguishes itself from other source end defenses, which use statistics to gather profiles. Information entropy, a measure to find correlation among traffic flows, is then used. Information entropy is used to deduce the current state of the dynamic network. Since the volume of the traffic at the source end would be moderate, it would be difficult to find the suspicious traffic at the source end. We found that the parameters we considered were good in identifying such traffic. We experimented our scheme using network simulator with network traffic traces and found the results were promising and presented them.

Behaviour of Reinforced Concrete Beams with SIFCON at Various Locations in the Beam

Abstract: This paper reports the flexural behaviour of conventional reinforced concrete beams with Slurry Infiltrated Fibrous Concrete (SIFCON) at various locations in the beam. Composite beams are cast with a layer of SIFCON at various locations of the conventional RC beam. Beams made entirely with RC, Fibre Reinforced Concrete (FRC) and SIFCON are also investigated for comparison. A total number of twenty one specimens corresponding to seven test series are cast and tested under two point flexural forward cyclic loading system. The parameters like load carrying capacity, stiffness degradation, ductility and energy absorption capacity are assessed. The concrete mix for RC beams are designed to obtain a concrete grade of M30. For SIFCON, the fibre volume fraction is 9%. The steel fibers used in the study are round crimped fibers having 0.5 mm diameter and aspect ratio of 60. Results indicate that the use of SIFCON in conventional RC beams have enhanced the cracking behavior in terms of considerable increase in the first crack load and in the formation of larger number of finer cracks. The ultimate load carrying capacity, stiffness, ductility and energy absorption capacity are increased to a greater extent for composite beams compared with that of conventional RC beams.