Robotic vehicles (RVs), such as drones and ground rovers, are a type of cyber-physical systems that operate in the physical world under control computing components cyber world. Despite RVs' robustness against natural disturbances, or attacks RVs may lead to malfunction subsequently disruption failure vehicles' missions. To avoid mitigate consequences, it is essential develop attack detection techniques for RVs. In this paper, we present novel framework identify external, on fly by deriving...

Real-time microcontrollers have been widely adopted in cyber-physical systems that require both real-time and security guarantees.Unfortunately, is sometimes traded for performance such systems.Notably, memory isolation, which one of the most established features modern computer systems, typically not available many microcontroller due to its negative impacts on violation constraints.As such, space these has created an open, monolithic attack surface attackers can target subvert entire...

Many applications that make use of sensor networks require secure communication. Because asymmetric-key solutions are difficult to implement in such a resource-constrained environment, symmetric-key methods coupled with priori key distribution schemes have been proposed achieve the goals data secrecy and integrity. These approaches typically assume all sensors similar terms capabilities, hence deploy same number keys network provide aforementioned protections. In this paper we demonstrate...

Topological features such as persistence diagrams and their functional approximations like images (PIs) have been showing substantial promise for machine learning computer vision applications. This is greatly attributed to the robustness topological representations provide against different types of physical nuisance variables seen in real-world data, view-point, illumination, more. However, key bottlenecks large scale adoption are computational expenditure difficulty incorporating them a...

Deep-learning architectures for classification problems involve the cross-entropy loss sometimes assisted with auxiliary functions like center loss, contrastive and triplet loss. These facilitate better discrimination between different classes of interest. However, recent studies hint at fact that these do not take into account intrinsic angular distribution exhibited by low-level high-level feature representations. This results in less compactness samples from same class unclear boundary...

This paper presents a method of prejoining the expected next TV channels for reducing IPTV channel zapping time with consideration surfing behavior and particular preference each viewer. In this method, home gateway multicast protocol proxy set-top boxes (STB) prejoins preferred based on aggregated data from STBs. The experimental results show that improves prediction accuracy prejoin little network bandwidth overhead, which leads to reduce services.

ARM is the leading processor architecture in emerging mobile and embedded market. Unfortunately, there has been a myriad of security issues on both systems. While many countermeasures such have proposed recent years, majority applications still cannot be patched or protected due to run-time space overhead constraints unavailability source code. More importantly, rapidly evolving market makes any platform-specific solution ineffective. In this paper, we propose RevARM, binary rewriting...

Mixup is a popular data augmentation technique based on creating new samples by linear interpolation between two given samples, to improve both the generalization and robustness of trained model. Knowledge distillation (KD), other hand, widely used for model compression transfer learning, which involves using larger network's implicit knowledge guide learning smaller network. At first glance, these techniques seem very different, however, we found that "smoothness" connecting link also...

In data-parallel synchronous training of deep neural networks, different devices (replicas) run the same program with partitions batch, but weight update computation is repeated on all replicas, because weights do not have a batch dimension to partition. This can be bottleneck for performance and scalability in typical language models large weights, small per-replica size which large-scale training. paper presents an approach automatically shard across replicas efficient communication...

We propose a new type of vulnerability for Robotic Vehicles (RVs), called Cyber-Physical Inconsistency. These vulnerabilities target safety checks in RVs (e.g., crash detection). They can be exploited by setting up malicious environment conditions such as placing an obstacle with certain weight and angle the RV's trajectory. Once exploited, may fail to report real physical accidents or false alarms (while RV is still operating normally). Both situations could lead life-threatening...

Activity recognition has been receiving significant attention from a variety of research areas such as human performance enhancement, health promotion, and computer interaction. However, recognizing activities accelerometer data still remains challenging problem due to sensitivity sampling rates, misalignment data, increased variability in among clinically relevant populations. In order solve these issues, we adopt methods functional analysis, which consider non-elastic rate variations...

Multi-sensor fusion (MSF) is widely used in autonomous vehicles (AVs) for perception, particularly 3D object detection with camera and LiDAR sensors. The purpose of to capitalize on the advantages each modality while minimizing its weaknesses. Advanced deep neural network (DNN)-based techniques have demonstrated exceptional industry-leading performance. Due redundant information multiple modalities, MSF also recognized as a general defence strategy against adversarial attacks. In this paper,...

Converting wearable sensor data to actionable health insights has witnessed large interest in recent years. Deep learning methods have been utilized and achieved a lot of successes various applications involving wearables fields. However, unique issues related sensitivity variability between subjects, dependency on sampling-rate for analysis. To mitigate these issues, different type analysis using topological shown promise as well. Topological (TDA) captures robust features, such persistence...

Energy consumption as well performance should be considered when designing high-performance multicore processors. The energy consumed in the instruction cache accounts for a significant portion of total processor consumption. Therefore, energy-aware design techniques are essential In this paper, we propose new architecture, which is based on level-0 composed filter and victim together, proposed architecture reduces by reducing number accesses to level-1 cache. We evaluate using simulation...