Fiky Yosef Suratman

Modified K-means algorithm using timestamp initialization in sliding window to detect anomaly traffic

Traffic anomalies that occur on the network usually make authorized users cannot access properly. That because by an increased number of users at a time or due to the attack of botnet to the network. This research purpose a method to detect there is anomaly traffic or not. This research used K-Means algorithm as the detection algorithm that modified on determination of the centroid and the cluster initialization, where the cluster initialization was used Timestamp Initialization as applied which in the determination of the centroid and the cluster based on the incoming data point. Expected modified K-Means using Timestamp Initialization can eliminate the determination of K-cluster that affect detection rate and false positive rate when using different K-cluster. This research also used windowing technique to obtain a better efficient process to detect anomaly traffic.

DDoS detection using modified K-means clustering with chain initialization over landmark window

Denial-of-service is a common form of network attack that affect user access right by preventing legitimate user from accessing certain information, thus giving great, disadvantage to the user and service provider. This paper present a method of denial-of-service detection using clustering technique with k-means algorithm which available to be modified and developed in many possible way. K-means algorithm used in this paper is modified using chain initialization over landmark window approach to process large amount of data and the result evaluated with detection rate, accuracy, and false positive rate. This method has been proven effective in detecting denial-of-service traffic using DARPA 98 dataset with satisfying result.

Spectrum sensing of OFDM signals using GLRT detector with bootstrap approach

Among some of its duties, the cognitive radios main role is to do spectrum sensing in the surrounding radio environment. In order to do mapping and can determine the frequency of use vacant frequencies which can be optimized for the use of cognitive radio communication. Even to the worst conditions in which the detector system has no knowledge of the signal and noise. There are several techniques that can be used to perform spectrum sensing in this condition, among these techniques are simpler implementation of energy detection techniques. This technique has a weakness at noise conditions are unknown and uncertainty. In this condition, a technique that can be used and better than energy detection is GLRT detector. However, this detector requires choose threshold with empirically. This process has a problem when we move from one location to another, it is necessary to conduct empirical calculations again. And would face difficulties if we do detect a signal that is already active in a new place, so it is necessary to know the exact time the signal was not active. In this condition the bootstrap approach can help determine the threshold detector directly from the active signal is received. So the detector can gain threshold which is always updated with any condition, anytime and anywhere. The simulation result show that GLRT detector with bootstrap approach has a toughness in the face of uncertainty noise. Even able to exceed the performance of which is owned by the energy detector and GLRT. At condition SNR -5 dB and uncertainty noise 1 dB, GLRT with bootstrap approach improve probability of miss detection almost 0.030 than ordinary GLRT and 0.094 than energy detector.

Enhanced LSB Steganography with people detection as stego key generator

Technology used to hide secret message in a communication called Steganography. Secret message can be text, image or any file that can be converted into binary. This secret message inserted into cover file which can be in form of image, sound or video, basiccaly cover file must be bigger than the secret message in size. Many methods have been proposed on how to hide secret messages in a cover file. These method include spatial domain which work on bit instead of statistically on cover file. Least Significant Bit have been long known as the simplest steganography embedding method. In this research we proposed a method of LSB with enhanced technique to increase security. To extract secret message from cover file one should have particular stego key that describe the location of the message and how to reconstruct them. We utilize a people detection method as a stego key since most of video footage involving human figure, from personal video footage, movie video or security camera recording. While maintaining the appearance of the cover video from arousing suspiciousness, the size of secret message that can be embedded to be one point of consideration.