As an Ethernet-based communication protocol with timing determinism, TSN (time-sensitive networking) has become a well-recognized promising in-vehicle network solution for increasingly automated automobiles. To satisfy the reliability requirement of safety-critical applications, existing works towards fault-tolerant TSN trade too much bandwidth for redundancy, limiting their scope of applicability. Targeting mixed-critical traffic, which is widely found in practice, we define a shared fault-tolerant segment that is compatible with the TSN standard. It serves the critical flows when faults occur, to improve their reliability and serves the non-critical flows when otherwise, to improve their quality of service (QoS). On top of this, we propose a space-time redundancy scheduling algorithm, aiming to make the most efficient use of bandwidth, i.e., to fulfill both the reliability as well as hard real-time requirements, and maximize the QoS with the least bandwidth. In essence, we formulate a bi-objective design space exploration problem with hundreds of thousands of decision variables and solve it with customized heuristics. Experimental results show that compared to the state-of-the-art methods, our reported work increases the number of critical flows that can be accommodated on a resource-constrained network by 3 to 4 times, and achieve the highest QoS with an average reduction of 60.3% in bandwidth. As the first work along sharing of bandwidth between mixed-critical traffic in fault-tolerant TSN, this idea can be further pursued towards higher efficiency and may be applied in general autonomous systems.

Retention, with its powerful energy-saving effect and negligible switching delay, offers the advantages of Power Gating (PG) and Clock Gating (CG), and has gradually been adopted into the processor power management systems by both AMD and Intel. However, there has been no research on how to integrate retention to achieve energy-efficient task scheduling. Furthermore, although most commercial chips support per-cluster configurations to achieve energy saving, existing energy-efficient works often underutilize all cores in the cluster, resulting in numerous static power from unused cores. In this paper, we propose a task partitioning algorithm to optimize energy consumption among clusters. We further study energy-efficient scheduling of directed acyclic graph (DAG) tasks with arbitrary deadlines on multicore platforms, which is the first work to integrate the per-cluster retention to reduce power consumption by fully utilizing the used clusters. The experimental results verify that our approach can increase the retention time by 21.24% and reduce the static energy consumption in the idle state by 45.81% compared to retention-based work-conserving scheduling.

There is a demand of high bandwidth in the emerging real-time applications, such as autonomous vehicles, robotics, and industrial automation, where time-sensitive networking (TSN) is a promising solution. According to IEEE 802.1, a port in a TSN switch has eight prioritized FIFO (first-in first-out) queues, whose gates are opened or closed following a gate control list (GCL). Most of the existing works use one TT (time-triggered) queue for the hard real-time traffic, i.e., traffic flows with hard deadlines, which easily achieves timing determinism through GCL. Unfortunately, as a rigid mechanism, GCL is not able to handle timing jitter. In this work, we propose a hybrid strategy towards robust TSN. GCL is applied to only one queue named TT T1 for a small number of hard real-time flows with negligible jitter. The remaining flows with hard deadlines are allocated to a prioritized queue named TT T2 without GCL. Similarly, GCL is removed from all other queues handling AVB (audio-video-bridging) flows with soft deadlines and BE (best-effort) flows with no deadlines. Two analyses are proposed to obtain delay bounds for the TT T2 flows and periodic AVB flows, respectively, with interference from TT T1. Although safety is not compromised if the periodic AVB flows miss their deadlines, it is often desirable in practice to satisfy them for quality of service. In order to strike a balance, contention between the AVB queues is resolved with credit values on top of priorities. Experiments support that the delay bounds for the TT T2 and AVB flows are safe. In addition, changing the credit function can lead to different delay bounds of AVB flows, which is valuable for real-world configurations of TSN.

Controller Area Network with Flexible Data-rate (CAN FD) is beneficial for the in-vehicle communication of Internet of Connected Vehicles (IoCVs) because of its high bandwidth and data field length. However, CAN FD lacks a security authentication mechanism, making it extremely vulnerable to masquerade attacks. This study proposes the security enhancement for a real-time parallel in-vehicle application adopting a two-stage method. The first stage obtains the lower bound of an in-vehicle application by quickly abandoning most of sequences, while the second stage enhances security by adding Message Authentication Codes (MACs) to messages taking advantage of the laxity interval from the lower bound to the deadline. Experiments with an example and the adaptive cruise control in-vehicle application show the advantage of the proposed two-stage method in increasing the total byte size of MACs.