Browsing by Author "Muhaidat, Sami"

Now showing 1 - 3 of 3
  • Thumbnail Image
    ItemOpen Access
    Cumulant-based automatic modulation classification over frequency-selective channels
    (IEEE, 2018-08) Yoo, Paul D.; Adly, Fatima; Muhaidat, Sami
    Automatic modulation classification (AMC), being an integral part of multi-standard communication systems, allows for the identification of modulation schemes of detected signals. The need for this type of blind modulation classification process can be evidently seen in areas such as interference identification and spectrum management. Consequently, AMC has been widely recognized as a key driving technology for military, security, and civilian applications for decades. A major challenge in AMC is the underlying frequency selectivity of the wireless channel, causing an increase in complexity of the classification process. Motivated by this practical concern, we propose the use of k-nearest neighbor (KNN) classifier based on higher-order of statistics (HOS), which are calculated as features to distinguish between different types of modulation types. The channel is assumed to b multipath frequency-selective and the modulation schemes considered are {2, 4, 8} phase-shift keying (PSK) and {16, 64, 256} quadrature amplitude modulation (QAM). The simulation results confirmed the superiority of this approach over existing methods.
  • Thumbnail Image
    ItemOpen Access
    Multi-Layered clustering for power consumption profiling in smart grids
    (IEEE, 2017-06-13) Al-Jarrah, Omar Y.; Al-Hammadi, Yousof; Yoo, Paul D.; Muhaidat, Sami
    Smart Grids (SGs) have many advantages over traditional power grids as they enhance the way electricity is generated, distributed, and consumed by adopting advanced sensing, communication and control functionalities that depend on power consumption profiles of consumers. Clustering algorithms (e.g., centralized clustering) are used for profiling individual’s power consumption. Due to the distributed nature and ever growing size of SGs, it is predicted that massive amounts of data will be created. However, conventional clustering algorithms neither efficient enough nor scalable enough to deal with such amount of data. In addition, the cost for transferring and analyzing large amounts of data is expensive high both computationally and communicationally. This paper thus proposes a power consumption profiling model based on two levels of clustering. At the first level, local power consumption profiles are derived, which are then used by the second level in order to create a global power consumption profile. The followed approach reduces the communication and computation complexity of the proposed two level model and improves the privacy of consumers. We point out that having good knowledge of the local power profiles leads to more effective prediction model and cost-effective power pricing scheme, especially in a heterogeneous grid topology. In addition, the correlations between the local and global profiles can be used to localize/identify power consumption outliers. Simulation results illustrate that the proposed model is effective in reducing the computational complexity without much affecting its accuracy. The reduction in computational complexity is about 52% and the reduction in the communicational complexity is about 95% when compared to the centralized clustering approach.
  • Thumbnail Image
    ItemOpen Access
    Semi-supervised multi-layered clustering model for intrusion detection
    (Elsevier, 2017-09-22) Al-Jarrah, Omar Y.; Al-Hammdi, Yousof; Yoo, Paul D.; Muhaidat, Sami; Al-Qutayri, Mahmoud
    A Machine Learning (ML) -based Intrusion Detection and Prevention System (IDPS) requires a large amount of labeled up-to-date training data, to effectively detect intrusions and generalize well to novel attacks. However, labeling of data is costly and becomes infeasible when dealing with big data, such as those generated by IoT (Internet of Things) -based applications. To this effect, building a ML model that learns from non- or partially-labeled data is of critical importance. This paper proposes a novel Semi-supervised Multi-Layered Clustering Model (SMLC) for network intrusion detection and prevention tasks. The SMLC has the capability to learn from partially labeled data while achieving a comparable detection performance to supervised ML-based IDPS. The performance of the SMLC is compared with well-known supervised ensemble ML models, namely, RandomForest, Bagging, and AdaboostM1 and a semi-supervised model (i.e., tri-training) on a benchmark network intrusion dataset, the Kyoto 2006+. Experimental results show that the SMLC outperforms all other models and can achieve better detection accuracy using only 20% labeled instances of the training data.