Hardware acceleration for verifiable, adaptive real-time communication (original) (raw)
A Fault-Tolerant Ethernet for Hard Real-Time Adaptive Systems
IEEE Transactions on Industrial Informatics
Distributed embedded systems (DESs) that perform critical tasks in unpredictable environments must be reliable, hard real-time, and adaptive. Since a DES comprises nodes that rely on a network, the network must provide adequate support: it must be reliable, convey messages on time, and meet new realtime requirements as the nodes adapt. Ethernet is ill-suited for such hard real-time adaptive systems, but it can be made suitable. The Flexible Time-Triggered (FTT) paradigm already supports hard real-time message exchanges and the necessary flexibility to meet evolving hard real-time requirements, but its Ethernet implementations had reliability limitations. To address these, we designed FTTRS, a communication subsystem that tolerates permanent and transient faults, even if they occur simultaneously, while keeping the paradigm's key features: support for both the timely exchange of periodic and sporadic real-time messages, and support for updating the real-time parameters of these messages at runtime. In this paper we present FTTRS, the first Ethernetbased communication subsystem specifically designed for highly reliable hard real-time adaptive DESs.
2003
This paper presents the design and implementation of RT-EP (Real-Time Ethernet Protocol), which is a softwarebased token-passing Ethernet protocol for multipoint communications in real-time applications, that does not require any modification to existing Ethernet hardware. This protocol allows the designer to model and analyze the real-time application using it, because it is based on fixed priorities and well-known schedulability analysis techniques can be applied. Furthermore, this protocol provides the applications the capacity of recovering from some fault conditions. It has been ported to an implementation of the Minimal Real-Time POSIX standard called MaRTE OS.
RT-EP: A Fixed-Priority Real Time Communication Protocol over Standard Ethernet
Lecture Notes in Computer Science, 2005
This paper presents the design and implementation of RT-EP (Real-Time Ethernet Protocol), which is a software-based token-passing Ethernet protocol for multipoint communications in real-time applications, that does not require any modification to existing Ethernet hardware. The protocol allows a fixed priority to be assigned to each message, and consequently well-known schedulability analysis techniques can be applied. A precise model of its timing behavior has been obtained. Furthermore, this protocol provides the ability of recovering from some fault conditions. It has been ported to an implementation of the Minimal Real-Time POSIX standard called MaRTE OS [10], and is being used to support real-time communications in an implementation of Ada's Distributed Systems Annex (RT-GLADE). It has been successfully used to implement a distributed controlled for an industrial robot.
A programmable arbitration layer for adaptive real-time systems
Adaptive real-time systems can respond to changes in the environment and thus allow for an extended range of oper-ations and for improved efficiency in the use of system resources. Building such adaptive real-time systems requires flexibility at each layer in the system stack. In this paper, we introduce and discuss our on-going effort to build a programmable arbitration layer. It builds on the Network Code language and enables the developer to program application-specific arbitration mechanisms which optimize bandwidth or encodes specific properties such as data redundancy, collision-free communication, and temporal isolation.
A new approach to provide real-time services on high-speed local area networks
Proceedings of the 15th …, 2001
In the past few years, networks of workstations (NOWs) and clusters, based on high-speed local area networks (LANs), have emerged as a serious alternative to supercomputers and high-performance servers. Meanwhile, applications demanding real-time network services have also suffered a substantial growth. In order to use NOWs for distributed real-time processing, a topology change and faulttolerant mechanism that guarantees the maximum latency or the minimum bandwidth in the worst case must be provided. Up to now, the backup channel protocol (BCP), based on real-time channels, provides fault-tolerant realtime services. But in this approach, fault tolerance is limited by the alternative paths provided by the routing function to establish the backup channels and topology change tolerance is not supported. On the other hand, dynamic reconfiguration updates the routing tables without stopping user traffic when a topology change or fault occurs. However, dynamic reconfiguration by itself does not provide neither quality of service nor real-time services, but it provides support for an additional mechanism designed to meet realtime requirements.
FPGA-based Implementation of an Ethernet Switch for Real-Time Applications
The use of switched Ethernet for precise and safe realtime communication still suffers from undesired phenomena, such as blocking caused by long non-preemptive frames, lack of protection against errors in the time domain, couplings across virtual LANs and priority levels via internal switch shared resources. Recently, a few solutions were proposed to cope with such phenomena. One such solution is based on an enhanced switch following the Flexible Time-Triggered paradigm, which enforces strict service differentiation, blocking-free forwarding and timing errors confinement. In this paper we propose a new architecture following an hardware-software co-design approach that facilitates the development of the enhanced switch features by separating the traffic scheduling from the common management activities associated to switching.
Deploying hard real-time control software on chip-multiprocessors
Proceedings - 16th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications, RTCSA 2010, 2010
Deploying real-time control systems software on multiprocessors requires distributing tasks on multiple processing nodes and coordinating their executions using a protocol. One such protocol is the discrete-event (DE) model of computation. In this paper, we investigate distributed discrete-event (DE) with null-message protocol (NMP) on a multicore system for real-time control software. We illustrate analytically and experimentally that even with the null-message deadlock avoidance scheme in the protocol, the system can deadlock due to inter-core message dependencies. We identify two central reasons for such deadlocks: 1) the lack of an upper-bound on packet transmission rates and processing capability, and 2) an unknown upper-bound on the communication network delay. To address these, we propose using architectural features such as timing control and real-time network-on-chips to prevent such message-dependent deadlocks. We employ these architectural techniques in conjunction with a distributed DE strategy called PTIDES for an illustrative car wash station example and later follow it with a more realistic tunnelling ball device application.
Comparing Admission Control Architectures for Real-Time Ethernet
IEEE Access, 2020
Industry 4.0 and Autonomous Driving are emerging resource-intensive distributed application domains that deal with open and evolving environments. These systems are subject to stringent resource, timing, and other non-functional constraints, as well as frequent reconfiguration. Thus, real-time behavior must not preclude operational flexibility. This combination is motivating ongoing efforts within the Time Sensitive Networking (TSN) standardization committee to define admission control mechanisms for Ethernet. Existing mechanisms in TSN, like those of AVB, its predecessor, follow a distributed architecture that favors scalability. Conversely, the new mechanisms envisaged for TSN (IEEE 802.1Qcc) follow a (partially) centralized architecture, favoring short reconfiguration latency. This paper shows the first quantitative comparison between distributed and centralized admission control architectures concerning reconfiguration latency. Here, we compare AVB against a dynamic real-time reconfigurable Ethernet technology with centralized management, namely HaRTES. Our experiments show a significantly lower latency using the centralized architecture. We also observe the dependence of the distributed architecture in the end nodes' performance and the benefit of having a protected channel for the admission control transactions.
Supporting Configurable Real-Time Communication Services
1997
Constructing communication services that provide real-ti me guarantees is important for many applications built on distributed systems. While a variety of such services have been designed and implemented, most are targeted for specific applications and are correspondingly difficult to adapt to differing requirements in other areas. This pape r presents an approach to building configurable and customized versions of real-time communication services based on software modules called micro-protocols. Each micro-protocol implements a different semantic property or property variant, and interacts with other micro-protocol s using an event-driven model supported by a runtime system providing real-time guarantees. The programming model is presented, together with an implementation design based on the x-kernel model for building network subsystems and the OSF/RI MK 7.2 operating system. The design of a highly-configurable real-time channel abstraction built using this approach is also given....