The Time-Triggered paradigm presents a lack of flexibility due to the required static scheduled. If an unpredicted event occurs, a new schedule needs to be synthesized. Centralized approaches have been proposed to obtain such schedules during runtime, while fully-distributed approaches seek to repair only the affected sections of the schedule. This paper proposes a Semi-Distributed Self-Healing Protocol that pursues to combine the benefits of both approaches. We study the applicability of our protocol repairing schedules after link failures. Early results show that link failures can be repaired in 2ms for the evaluated network.

Real-time networks play a fundamental role in embedded systems. To meet timing requirements, provide low jitter and bounded latency in such networks the time-triggered communication paradigm is frequently applied in such networks. In this paradigm, a schedule specifying the transmission times of all the traffic is synthesized a priori. Given the steady increase in size and complexity of embedded systems, coupled with the addition of wireless communication, a new time-triggered network model of larger and mixed wired-wireless network isdeveloping. Developing such next-generation networks entails significant research challenges, especially concerning scalability, i.e., allowing generation of schedules of the very large next-generation networks in a reasonable time. A second challenge concerns a well-known limitation of the time-triggered paradigm: its lack of flexibility. Large networks exacerbate this problem, as the number of changes during network operation increases with the number of components, which renders static scheduling approaches unsuitable.

In this thesis, we first propose a remedy to the scalability challenge that the synthesis of next-generation network schedules introduces. We propose a family of divide-and-conquer approaches that segment the entire scheduling problem into small enough subproblems that can be effectively and efficiently solved by state-of-the-art schedulers. Second, we investigate how adaptive behaviours can be introduced into the time-triggered paradigm with the implementation of a Self-Healing Protocol. This protocol addresses the flexibility challenge by only updating a small segment of the schedule in response to changes during runtime. This provides a significant advantage compared to current approaches that fully reschedule the network. In the course of our research, we found that our protocol become more effective when the slack in the original schedule is evenly distributed during the schedule synthesis. As a consequence, we also propose a new scheduling approach that maximizes the distances between frames, increasing the success rate of our protocol.

The divide-and-conquer approaches developed in this thesis were able to synthesize schedules of two orders of magnitude more traffic and one order of magnitude more nodes in less than four hours. Moreover, when applied to current industrial size networks, they reduced the synthesis time from half an hour to less than one minute compared with state-of-the-art schedulers. The Self-Healing Protocol opened a path towards adaptive time-triggered being able to heal schedules online after link and switch failures in less than ten milliseconds.

Future cyber–physical systems may extend over broad geographical areas, like cities or regions, thus, requiring the deployment of large real-time networks. A strategy to guarantee predictable communication over such networks is to synthesize an offline time-triggered communication schedule. However, this synthesis problem is computationally hard (NP-complete), and existing approaches do not scale satisfactorily to the required network sizes. This article presents a segmented offline synthesis method which substantially reduces this limitation, being able to generate time-triggered schedules for large hybrid (wired and wireless) networks. We also present a series of algorithms and optimizations that increase the performance and compactness of the obtained schedules while solving some of the problems inherent to segmented approaches. We evaluate our approach on a set of realistic large-size multi-hop networks, significantly larger than those considered in the existing literature. The results show that our segmentation reduces the synthesis time by up to two orders of magnitude.

The time-triggered communication paradigm has been shown to satisfy temporal isolation while providing end to end delay guarantees through the synthesis of an offline schedule. However, this paradigm has severe flexibility limitations as any unpredicted change not anticipated by the schedule, such as a component failure, might result in a loss of frames. A typical solution is to use redundancy or replace and update the schedule offline anew. With the ever increase in size of networks and the need to reduce costs, supplementary solutions that enhance the reliability of such networks are also desired. In this paper, we introduce a repair algorithm capable of reacting to unpredicted link failures. The algorithm quickly modifies the schedule such that all frames are transmitted again within their timing guarantees. We found that the success of our algorithm increases significantly with the existence of empty slots spread over the schedule, an opposite approach compared to packing frames, commonly used in the literature. We propose a new ILP formulation that includes a maximization of frame and link intermissions to stretch empty slots over the schedule. Our results show that we can repair with 90% success rate within milliseconds to a valid schedule compared to a few minutes needed to re-schedule the whole network. 

Adaptive requirements for networks with strict timing restrictions do challenge the static nature of the time-triggered communication paradigm. Continuous changes in the network topology during operation require frequent rescheduling, followed by schedule distribution, a process that is excessively time-consuming as it was intended to be performed only during the design phase. The fully-distributed Self-Healing Protocol introduced a collaborative method to quickly modify the local schedules of the nodes during runtime, after link failures. This protocol gets the network back to correct operation in milliseconds, but it assumes that only the nodes are able to modify their local schedules, which limited the achieved improvement. This paper proposes to shift to a semi-distributed strategy, where high-performance nodes are responsible for the nodes and links within a small network segment. These nodes rely on their privileged view of the system in order to reduce the response time, increase the healing success rate, and extend the fault model to include switch failures. 

Time-Triggered communication is based on generating an offfine static schedule that guarantees frame transmissions with reduced latency and low jitter. However, static schedules are not adaptive: if some unpredicted event happens, like a link failure, the schedule is not valid anymore and a new one needs to be synthesized from scratch. This paper presents a novel hot-patching protocol which seeks, after a link failure disconnecting two nodes, to find a new path to reconnect both nodes and restore during run-time the affected part of the schedule. We also introduce the concept of reparability as a desired property of the schedule, which increases the probability of our protocol to succeed. The first evaluation shows that our hot-patching protocol can recover from a link failure consistently in less than 25ms.

Wireless communications enables introduction of Internet of Things (IoT) in industrial networks. Unfortunately, real-time guarantees required for many IoT applications, may be compromised in wireless networks due to an unreliable transmission medium. A key component in enabling real-time communications is the medium access control (MAC) layer and its ability to effectively avoid concurrent transmissions that causes deadline misses. Also, deploying the network in a harsh interference environment can lead to low reliability. Time diversity, based on transmitting several copies of the same data at different instants, increases reliability but at the expense of increased jitter and bandwidth. A more efficient resource utilization is expected from cognitive radio, which dynamically takes into account the status of the wireless environment before performing transmissions. This paper proposes a wireless MAC protocol based on scheduled timeslots to avoid concurrent transmissions, combined with two different mechanisms to increase reliability, one based on time diversity and another on cognitive radio. The protocol and its mechanisms to enhance reliability are compared in different interference scenarios, and show that cognitive radios achieves better performance than time diversity, especially when the interference is produced by a jammer.

For handling frame transmissions in highly deterministic real-time networks, i.e. networks requiring low communication latency and minimal jitter, an offline time-triggered schedule indicating the dispatch times of all frames can be used. Generation of such an offline schedule is known to be a NPcomplete problem, with complexity driven by the size of the network, the number and complexity of the traffic temporal constraints, and link diversity (for instance, coexistence of wired and wireless links). As embedded applications become more complex and extend over larger geographical areas, there is a need to deploy larger real-time networks, but existing schedule synthesis mechanisms do not scale satisfactorily to the sizes of these networks, constituting a potential bottleneck for system designers. In this paper, we present an offline synthesis tool that overcomes this limitation and is capable of generating time-triggered schedules for networks with hundreds of nodes and thousands of temporal constraints, also for systems where wired and wireless links are combined. This tool models the problem with linear arithmetic constraints and solves them using a Satisfiability Modulo Theory (SMT) solver, a powerful general purpose tool successfully used in the past for synthesizing time-triggered schedules. To cope with complexity, our algorithm implements a segmented approach that divides the total problem into easily solvable smaller-size scheduling problems, whose solutions can be combined for achieving the final schedule. The paper also discusses a number of optimizations that increase the size and compactness of the solvable schedules. We evaluate our approach on a set of realistic large-size multi-hop networks, significantly bigger than those in the existing literature. The results show that our segmentation reduces the synthesis time dramatically, allowing generation of extremely large compact schedules.

Many embedded systems with real-time requirements demand minimal jitter and low communication end-to-end latency for its communication networks. The time-triggered paradigm, adopted by many real-time protocols, was designed to cope with these demands. A cost-efficient way to implement this paradigm is to synthesize a static schedule that indicates the transmission times of all the time-triggered frames such that all requirements are met. Synthesizing this schedule can be seen as a bin-packing problem, known to be NPcomplete, with complexity driven by the number of frames. In the last years, requirements on the amount of data being transmitted and the scalability of the network have increased. A solution was proposed, adapting real-time switched Ethernet to benefit from its high bandwidth. However, it added more complexity in computing the schedule, since every frame is distributed over multiple links. Tools like Satisfiability Modulo Theory solvers were able to cope with the added complexity and synthesize schedules of industrial size networks. Despite the success of such tools, applications are appearing requiring embedded systems with even more complex networks. In the future, real-time embedded systems, such as large factory automation or smart cities, will need extremely large hybrid networks, combining wired and wireless communication, with schedules that cannot be synthesized with current tools in a reasonable amount of time. With this in mind, the first thesis goal is to identify the performance limits of Satisfiability Modulo Theory solvers in schedule synthesis. Given these limitations, the next step is to define and develop a divide and conquer approach for decomposing the entire scheduling problem in smaller and easy solvable subproblems. However, there are constraints that relate frames from different subproblems. These constraints need to be treated differently and taken into account at the start of every subproblem. The third thesis goal is to develop an approach that is able to synthesize schedules when different frame constraints related to different subproblems are inter-dependent. Last, is to define the requirements that the integration of wireless communication in hybrid networks will bring to the schedule synthesis and how to cope with the increased complexity. We demonstrate the viability of our approaches by means of evaluations, showing that our method is capable to synthesize schedules of hundred of thousands of frames in less than 5 hours.

Time-triggered offline scheduling is a cost-efficien way to guarantee low communication end-to-end latency and minimal jitter for communication networks in real-time systems. The schedule is generated pre-runtime and indicates the transmission times of time-triggered frames such that contention is prevented. The synthesis of such offline schedules is a bin-packing problem, known to be NP-complete, with complexity driven by the constraints on frame transmissions, and the number of frames in the schedule. Satisfiability Modulo Theories combined with segmented approaches have been successfully used for synthesizing schedules of large networks. However, such synthesis did not take into account frames periods that are much shorter than the time to execute the schedule cycle. This paper presents a periodaware segmented approach that takes into account the frame periods in order to allocate various instances of a frame within a single cycle. We describe three different synthesis strategies and evaluate them with different synthetic experiments. The results show better performance for one of the strategies, which can synthesize schedules of large networks with high communication loads in less than one hour. We also report how the synthesis time and the schedule quality can change with different parameter configurations.