Chroma tools

This study proposes a non-contact driver monitoring system based on millimeter-wave FMCW radar. It integrates temporal convolution, spatial attention, and a cross-gated fusion mechanism in a multi-modal deep learning model called RaTSNet to recognize driver behaviors and estimate heart and respiratory rates in real time. The system features an adaptive sensing mechanism that adjusts the monitoring range based on driver position and signal quality, enhancing the precision of physiological signal detection. Deployed on the NVIDIA Jetson Orin Nano platform, the system achieves over 94% accuracy in detecting high-risk driving behaviors, demonstrating strong practical potential.

Driver maneuver prediction is of great importance in designing a modern Advanced Driver Assistance System (ADAS). Such predictions can improve driving safety by alerting the driver to the danger of unsafe or risky traffic situations. In this research, we developed a model to predict driver maneuvers, including left/right lane changes, left/right turns and driving straight forward 3.6 seconds on average before they occur in real time. For this, we propose a deep learning method based on Long Short-Term Memory (LSTM) which utilizes data on the driver's gaze and head position as well as vehicle dynamics data. We applied our approach on real data collected during drives in an urban environment in an instrumented vehicle. In comparison with previous IOHMM techniques that predicted three maneuvers including left/right turns and driving straight, our prediction model is able to anticipate two more maneuvers. In addition to this, our experimental results show that our model using identical dataset improved F1 score by 4% and increased to 84%.

To avoid traffic accidents caused by driver's fatigue, this paper designs a fatigue detection system based on ADSP-BF548 and relative algorithm. It captures driver's images in real time by on-board camera and localizes driver's eye position via frame-difference projection, edge extracting, and template matching such kind of methods. At last, makes a warning in time according to the eye states in a period of time. The experiment shows this driver fatigue detection system based on ADSP-BF548 is of high accuracy.

The paper discusses a software solution for contention between Long Term Evolution (LTE) layer 1 (L1) Time Division Duplex (TDD) software which is a soft real time system with strict timing compliance requirements and Linux Real Time operating system over shared Serial Peripheral Interface (SPI). The cores of the asymmetric multiprocessor systems are agnostic to each other's activities and share SPI in real time with no synchronization mechanism existing between them. The solution effectively mitigates the possibility of contention by converting non deterministic SPI transactions into externally triggered deterministic SPI transactions and provides mechanism to detect contention if it occurs under high L1 load scenarios. In the case of contention, an effective retry mechanism is put in place as a part of proposed solution. The paper concludes by providing various scenarios in which the proposed solution can be leveraged.

Real time DDoS detection using fuzzy estimators

We use an analytical formulation based on Markov chains to evaluate blocking and energy-saving performance of a Wavelength Division Multiplexing (WDM) network with transponder power management. We consider three power modes for transponders, i.e., powered-on (ON), idle (sleep) or powered-off (OFF). To guarantee fast provisioning to high priority (e.g., real time) connections, we assume that some idle transponders (that can be re-activated much faster than powered-off transponders) are reserved for such connections. Some other connections may not have strict set-up time constraint, and can wait until a powered-off transponder is activated and available. We find that at least 32% of power savings can be obtained with such management strategy, for an acceptable degradation in the overall network blocking performance.

Real Time Fast Ultrasound Imaging Technology and Possible Applications

Industrial robots are complex systems with strict real time, reliability, availability, and safety requirements. Robot controllers are the basic components of the product-line architecture of these systems. They are complex real time computers which control the mechanical arms of a robot. By their nature, robot controllers are generic and open computer systems, because to be useful, they must be programmable by end-users. This is typically done by using software configuration parameters and a domain and vendor-specific programming language. For some purposes, this may not be sufficient. A means of adding low-level software extensions to the robot controller, basically extending its base software platform is needed when, for example, a third party wants to add a completely new sensor type that is not supported by the platform. Any software platform evolution in this direction introduces a new set of broad quality issues and other concerns. Dependability concerns, especially safety, reliability and availability, are among the most important for robot systems. In this paper, we use the ABB robot controller to show how an architecture transformation approach based on quality attributes can be used in the design process for increasing the platf openness.

Bountiful reasons are there for crop up of road accidents. Predominantly most of the road accidents are turn out by the humans. In compare with other sources of road accidents, distracted driving is the foremost one. Driver gets distracted ascribed to many factors such as texting using phone, seeing outside of the car. The regime can grasp the road accidents by enforcing the traffic regulations moreover the regime seeks the assistance from the technology to reduce the car accidents due to driver distraction. There are numerous research undergoing to reduce the chance of the driver to get distracted by monitoring the physical activities of the driver while driving the four wheeler. Automatic detection of driver distraction can support to develop an apprise system by observing the activities of the driver that provide better result and evading the car accidents. Deep learning is a leading technology in automated industries to detect the traffic light signs and pedestrians. The novel ResNext 101 is the proposed work and inherited from its parent model called Residual Network (ResNet) to train and test the model. The “Next” in the word ResNext represents next dimension of ResNet called as “cardinality”. In this paper, with the help of Convolutional Neural Network (CNN), the system detects and captures the distracted moments of the driver from proper driving gestures automatically in the form of 2D image. The test results of the model outperforms and proved its significance over the existing driver detection algorithms by bagging higher efficacy with 97.6% accuracy in classifying the live driver posture.

We present a method for providing response time guarantees in Asynchronous Transfer Mode (ATM) networks. The method is based on traditional real time CPU Response Time Analysis (RTA), and is intended to be used for admission control of hard real time traffic. The method determines if a new connection can be admitted without violating the strict timing requirements specified for the new as well as old connections. We illustrate the merits of our method by comparing it with Weighted Fair Queuing (WFQ) and the Calculus for Network Delays (CND). Two types of comparisons are made. In the first, we evaluate how well the associated analysis can accommodate different traffic scenarios and loads, and in the second comparison we use simulation to compare observed worst case behaviors with estimates obtained by the analysis. The comparisons clearly indicate that RTA outperforms both WFQ and CND for a set of realistic traffic scenarios.

Our last decade has experienced significant growth in terms of data generated by billions of people connected to the Internet. Recent prognoses about Big Data, Internet of Thing, Cloud Computing show growing demand for an efficient processing of huge amount of data with strict time limits. The Best data distribution on Shared Nothing Architecture (SN) is a major issue. SDDS (Scalable Distributed Data Structure) is one of the most important data structure family that is able to store, manage a huge amount of data in distributed environments. In this paper, we present an SDDS driven approach in which data is partitioned using the dynamic intervals. We are targeting any application that handles a large volume of data to be distributed over a cluster of parallel processing. This volume of data is continuously supplied in real time from multiple sources. In such situation, to face the risk of major data imbalances (Data Skew) that would influence negatively the performance of parallel processing (such as range queries), we propose a self-balancing range partitioning scheme allowing nodes processing to dynamically adjust the boundaries of their ranges. Experimental results show that our approach significantly improves processing time, specially for data sorting operation, keep data balanced over the cluster.

The main objective of this research is to find the best evacuation path during a flood disaster emergency situation. The challenge lies in how to manage this situation that has dynamic changing conditions and strict time constraints. In this paper, we propose an adaptive system for flood evacuation. This system gives the best decisions to take, in this emergency situation, to minimize damages. It computes the best evacuation routes in real time by executing an algorithm which takes into consideration the spatial characteristics of hazard propagation. The system, based on sensors network, is adapted to the changes of the environment and provides directions for the shortest and less hazardous routes to the users.

Real time extraction of tracking error signal in IR seeker with a plus stationary reticle

The picky needs raised in the embedded market lead to the growth of real-time operating systems (RTOS), which differ from the general-purpose operating systems (GPOS) by applying strict timing and providing predictable execution of tasks. Some features of a real-time application, such as human interaction and some data transfers to other components are not time-critical. As the general-purpose operating system requires access to the data gathered by the real-time operating system, these two should be consolidated on the same platform form a hybrid system. This proposal examines how the existing Sloth approach of interrupt-driven scheduling can apply to a hybrid system running both a real-time and the Linux kernel simultaneously on the unchanged hardware. This entails a make another study of existing approaches towards a real-time capable Linux and the design of a communication channel between the real-time core and processes running in Linux kernel. The evaluation of the resulting Slothful Linux is supposed to show the advantages of the Sloth approach in a hybrid system compared to both the standalone Sloth and another real-time conservatory for Linux.

Internet was based on simple principles, providing only one best-effort service, i.e. without any delay or packet loss guarantee. This service is sufficient for applications such as electronic mail which have no real time constraints. However, video communication needs to be monitored and adapted when it traverses best-effort environment. The transmission of a continuous flow, like streaming video, witness very strict time constraints. That’s why we need services adapted to specific application needs with a guaranteed QoS (Quality of Service). This paper presents a context-aware mechanism for the improvement of the wireless video applications quality. This can be done by negotiating the desired QoS between the user terminal and the network entities and by adapting the video communication's rate to the channel's current situation. The decision regarding the needed QoS parameters and the rate control is based on context information (user’s profile, applications needs and network monitoring).