Nonvolatile memory express (NVMe) solid-state drives (SSDs) have become mainstream storage devices in today’s computing systems, due to their high throughput and ultralow latency. It has been observed that the impact of interference among all concurrently running streams (i.e., I/O workloads) on their overall responsiveness differs significantly in multistream SSDs, resulting in unfairness. This article proposes a cache division management scheme built on top of the evenly partition scheme for NVMe SSDs, to enhance I/O responsiveness without consciously sacrificing fairness. To this end, we first build a mathematical model to directly cut portions from the Local cache partitions allocated to concurrently running streams, considering their run-time performance measures. Then, our approach pools these portions together for the use of all streams. As a result, each stream has its corresponding Local cache space for ensuring fairness, meanwhile the pooled Global cache space is shared by all streams for enhancing I/O responsiveness. Trace-driven simulation experiments demonstrate that our proposal reduces the overall I/O latency by up to24.4%, and improve the measure of fairness by$\mathtt{2.5}\times $on average, in contrast to existing cache management schemes for NVMe SSDs.

Small object detection has a broad application prospect in image processing of unmanned aerial vehicles, autopilot and remote sensing. However, some difficulties exactly exist in small object detection, such as aggregation, occlusion and insufficient feature extraction, resulting in a great challenge for small object detection. In this paper, we propose an improved algorithm for small object detection to address these issues. By using the spatial pyramid to extract multi-scale spatial features and by applying the multi-scale channel attention to capture the global and local semantic features, the spatial pooling pyramid and multi-scale channel attention module (SPP-MSCAM) is constructed. More importantly, the fusion of the shallower layer with higher resolution and a deeper layer with more semantic information is introduced to the neck structure for improving the sensitivity of small object features. A large number of experiments on the VisDrone2019 dataset and the NWPU VHR-10 dataset show that the proposed method significantly improves the Precision, mAP and mAP50 compared to the YOLOv5 method. Meanwhile, it still preserves a considerable real-time performance. Undoubtedly, the improved network proposed in this paper can effectively alleviate the difficulties of aggregation, occlusion and insufficient feature extraction in small object detection, which would be helpful for its potential applications in the future.

Most solid-state drives (SSDs) adopt an on-board Dynamic Random Access Memory (DRAM) to buffer the write data, which can significantly reduce the amount of write operations committed to the flash array of SSD if data exhibits locality in write operations. This article focuses on efficiently managing the small amount of DRAM cache inside SSDs. The basic idea is to employ the visibility graph technique to unify both temporal and spatial locality of references of I/O accesses, for directing cache management in SSDs. Specifically, we propose to adaptively generate the visibility graph of cached data pages and then support batch adjustment of adjacent or nearby (hot) cached data pages by referring to the connection situations in the visibility graph. In addition, we propose to evict the buffered data pages in batches by also referring to the connection situations, to maximize the internal flushing parallelism of SSD devices without worsening I/O congestion. The trace-driven simulation experiments show that our proposal can yield improvements on cache hits by between 0.8 % and 19.8 %, and the overall I/O latency by 25.6 % on average, compared to state-of-the-art cache management schemes inside SSDs.

Earliest Deadline First (EDF) is the most widely studied optimal dynamic scheduling algorithm for uniprocessor real-time systems. In the existing literature, however, there is no complete exact analysis for EDF scheduling when both resource sharing and release jitter are considered. Since resource sharing and release jitter are important characteristics of embedded real-time systems, a solid theoretical foundation should be provided for EDF scheduled systems. In this paper, we extend traditional processor demand analysis to let arbitrary deadline real-time tasks share non-preemptable resources and suffer release jitter. A complete and exact schedulability analysis for EDF scheduled systems is provided. This analysis is incorporated into QPA (Quick Processor-demand Analysis) which provides an efficient implementation of the exact test.

The most important character of real-time systems is that they have stringent timing deadlines that must be guaranteed. A hard real-time system is required to complete its operations before all its timing deadlines. For a given task set, it is useful in an engineering context to know what changes to period can be made to a task that will deliver a schedulable system. In this paper, we develop the sensitivity analysis of task period for EDF scheduled systems on a uniprocessor. We prove that a minimum task period can be determined by a single pass of the QPA algorithm, an improved scheme is presented by using different initial values of the period. The approaches developed for sensitivity analysis of task period are therefore as efficient as QPA, and are easily incorporated into a system design support tool.

The correctness of a real-time system depends on not only the system's output but also on the time at which results are produced. A hard real-time system is required to complete its operations before all its timing deadlines. For a given task set it is useful to know what is the minimum speed of the processor that will deliver a schedulable system. It is also beneficial in an engineering context to know what changes to computation time can be made to a task that will result in a system that is borderline schedulable. In this paper, we address the sensitivity analysis (parameter calculations) for task execution times and speed of the processor for EDF-scheduled systems on a uniprocessor. We prove that an optimal (minimum or maximum) task parameter can be determined by a single pass of the QPA algorithm. This algorithm provides efficient and exact sensitivity analysis for arbitrary deadline real-time systems. The approaches developed for task parameter computations are therefore as efficient as QPA, and are easily incorporated into a system design support tool.

Earliest Deadline First (EDF) is an optimal scheduling algorithm for uniprocessor real-time systems. Quick Processor-demand Analysis (QPA) provides efficient and exact schedulability tests for EDF scheduling with arbitrary relative deadline. In this paper, we propose Improved Quick Processor-demand Analysis (QPA*) which is based on QPA. By extensive experiments, we show that QPA* can significantly reduce the required calculations to perform an exact test for unschedulable systems. We prove that the computation time for testing schedulable systems is hardly affected. Hence the required calculations for general systems can be significantly decreased.

Real-time scheduling is the theoretical basis of real-time systems engineering. Earliest deadline first (EDF) is an optimal scheduling algorithm for uniprocessor real-time systems. Existing results on an exact schedulability test for EDF task systems with arbitrary relative deadlines need to calculate the processor demand of the task set at every absolute deadline to check if there is an overflow in a specified time interval. The resulting large number of calculations severely restricts the use of EDF in practice. In this paper, we propose new results on necessary and sufficient schedulability analysis for EDF scheduling; the new results reduce, exponentially, the calculation times, in all situations, for schedulable task sets, and in most situations, for unschedulable task sets. For example, a 16-task system that in the previous analysis had to check 858,331 points (deadlines) can, with the new analysis, be checked at just 12 points. There are no restrictions on the new results: each task can be periodic or sporadic, with relative deadline, which can be less than, equal to, or greater than its period, and task parameters can range over many orders of magnitude.

Nowadays various kinds of anomalies are prohibiting the widely used Internet from offering normal services. Within them a novel anomaly is caused by bandwidth attacks. To defense these threats many detecting schemes are essentially based on unidirectional checking of traffic changes. When legitimately abrupt changes appear, they might result in false alarms. In this paper we consider the problem from the bidirectional-traffic view and analyze the traffic characteristics by checking the input/output traffic characteristics of the protected network node. We have analyzed the relationship between input and output traffic volume pairs in the simulation traffic and studied them both under normal and abnormal cases. Based on these analyses, we've proposed a heuristic DDoS flooding attack detection method and showed a verifying simulation as well.

This paper focuses on scheduling different hard real-time applications on a uniprocessor when the earliest deadline first algorithm is used as the local scheduler, and the global scheduler of the system could be fixed priority (FP) or earliest deadline first (EDF). Each application task could be periodic or sporadic, bound or unbound, with arbitrary relative deadline which could be less than, equal to or greater than its period. A number of different server types are considered. This paper presents an exact and efficient schedulability test for the application tasks based on the capacity demand criterion when the global scheduler could be FP or EDF, in some cases, it is necessary and sufficient. Schedulability tests which are necessary and sufficient for several types of dynamic servers are presented when the global scheduler is EDF.