No abstract

Most of the existing research works on the intrusion detection problem in a wireless sensor network (WSN) assume linear or random mobility patterns in abstracting intruders' models in traversing the WSN field. However, in real-life WSN applications, an intruder is usually an intelligent mobile robot with environment learning and detection avoidance capability (i.e., the capability to avoid surrounding sensors). Due to this, the literature results based on the linear or random mobility models may not be applied to the real-life WSN design and deployment for efficient and effective intrusion detection in practice. This motivates us to investigate the impact of intruder's intelligence on the intrusion detection problem in a WSN for various applications. To be specific, we propose two intrusion algorithms, the pinball and flood-fill algorithms, to mimic the intelligent motion and behaviors of a mobile intruder in detecting and circumventing nearby sensors for detection avoidance while heading for its destination. The two proposed algorithms are integrated into a WSN framework for intrusion detection analysis in various circumstances. Monte Carlo simulations are conducted, and the results indicate that: (1) the performance of a WSN drastically changes as a result of the intruder's intelligence in avoiding sensor detections and intrusion algorithms; (2) network parameters, including node density, sensing range and communication range, play a crucial part in the effectiveness of the intruder's intrusion algorithms; and (3) it is imperative to integrate intruder's intelligence in the WSN research for intruder detection problems under various application circumstances.

We present a new parallel algorithm for probabilistic graphical model optimization. The algorithm relies on data-parallel primitives (DPPs), which provide portable performance over hardware architecture. We evaluate results on CPUs and GPUs for an image segmentation problem. Compared to a serial baseline, we observe runtime speedups of up to 13X (CPU) and 44X (GPU). We also compare our performance to a reference, OpenMP-based algorithm, and find speedups of up to 7X (CPU). arXiv:1809.05018v1 [cs.DC] 13 Sep 2018 2 microct.lbl.gov 3 Cori configuration page: http://www.nersc.gov/users/ computational-systems/cori/configuration/ 4 Edison configuration page: http://www.nersc.gov/users/ computational-systems/edison/configuration/

No abstract

No abstract