Fault-tolerant real-time systems for emerging critical applications like wearable electronic healthcare monitors, consumer-grade unmanned aerial vehicles, or environmental monitoring have to tolerate errors during operation. If they fail, the consequences are dire. But often their budgets for error mitigation capabilities are low, which requires to limit mitigation capabilities to the most critical functions of a system. Still, less critical functions of a system should operate as long as possible, but current state-of-the-art scheduling approaches either ignore them or suffer from low acceptance rates. Our fully static and verification friendly mixed-criticality approach guarantees mentioned systems that the most critical system functions are always available, and maximizes the time where less critical system functions are operational: We prove that our approach is feasible, and extends the system operation time while providing full service by a factor of 1.93 with a probability of 0.92. With 1.56 higher acceptance rates compared to similar state-of-the-art approaches, the integration of functionalities of different criticality in one fault-tolerant system succeeds more and more often, which especially benefits emerging critical applications with limited budgets for error mitigation.

Mixed-criticality scheduling in modern deeply embedded mission and safety-critical systems needs to consider delivered service, that is, the runtime in low criticality mode. If the change into a higher criticality mode is triggered by the first overrunning job, the service is severely reduced. With earliest deadline first with virtual deadlines for single errors (EDF-VD-SE) we show how to reserve additional time to tolerate a single overrunning job by formulating and solving an optimization problem, and that EDF-VD-SE is feasible without assumptions about error probabilities for safety guarantees. We conduct extensive simulation experiments to report on average doubled service figures, and show how EDF-VD-SE results in a nearly constant acceptance rate of random task systems.

Correct scheduling in hard real-time systems is of utmost importance to guarantee that deadlines are not missed. By mathematical proof, correctness can be demonstrated for the worst case. Such demonstrations usually do not consider errors during system run-time, and do not provide quality of service insights. Such insights can be derived from simulations, but typical simulators are too slow for long term simulations.We developed Thready, a fast simulator for sporadic task systems under errors to investigate the long term behavior of scheduled systems. Thready’s three order of magnitude speedup in latency compared with the fastest state of the art simulator framework allows designers to investigate system performance in the average case, which facilitates understanding and better design decisions.

Network-on-chip (NoC) is the most promising design paradigm for the interconnect architecture of a multiprocessor system-on-chip (MPSoC). On the downside, a NoC has a significant impact on the overall energy consumption of the system. NoC simulators are highly relevant for design space exploration even at an early stage. Since links in NoC consume up to 50% of the energy, a realistic energy consumption of links in NoC simulators is important. This work presents a simulation environment which implements a technique to precisely estimate the data dependent link energy consumption in NoCs with virtual channels for the first time. Our model works at a high level of abstraction, making it feasible to estimate the energy requirements at an early design stage. Additionally, it enables the fast evaluation and early exploration of low-power coding techniques. The presented model is applicable for 2D and 3D NoCs. A case study for an image processing application shows that the current link model leads to an underestimate of the link energy consumption by up to a factor of four. In contrast, the technique presented in this paper estimates the energy quantities precisely with an error below 1% compared to results obtained by precise, but computational extensive, bit-level simulation.

Network-on-chip (NoC) is the most promising design paradigm for the interconnect architecture of a multiprocessor system-on-chip (MPSoC). On the downside, a NoC has a significant impact on the overall energy consumption of the system. This work presents the first technique to precisely estimate the data dependent link energy consumption in NoCs with virtual channels. Our model works at a high level of abstraction, making it feasible to estimate the energy requirements at an early design stage. Additionally, it enables the fast evaluation and early exploration of low-power coding techniques. The presented model is applicable for 2D as well as 3D NoCs. A case study for an image processing application shows that the current link model leads to an underestimate of the link energy consumption by up to a factor of four. In contrast, the technique presented in this paper estimates the energy quantities precisely (error below 1 %).

Through-silicon vias (TSVs) in 3D ICs show a significant power consumption, which can be reduced using coding techniques. This work presents an approach which reduces the TSV power consumption by a signal-aware bit assignment which includes inversions to exploit the MOS effect. The approach causes no overhead and results in a guaranteed reduction of the overall power consumption. An analysis of our technique shows a reduction in the TSV power consumption by up to 48 % for real correlated data streams (e.g. image sensor), and 11 % for low-power encoded random data streams.

The probability that a particular device is operational for a given duration, or reliability, is a dependability attribute and key metric for systems in critical applications. For example, systems for long-term autonomous exploration missions have to be operational during their complete mission. Other critical applications like banking, medical automotive or aerospace face similar reliability requirements that are only met by dependable systems. Traditional dependable systems, compared to their non-dependable counterparts, have three key issues: They are more expensive, consume more power, and provide less performance.

Extended monitoring of housekeeping data is required to increase the observability of a spacecrafts health status, its environment and resulting mechanical stress as well as physical parameters like the spacecrafts position and orientation. This implies the application of an increasing number of onboard sensors for various physical quantities like temperature, vibration, acceleration, voltage, current and others. These sensors need to offer high resolution in the time domain and high accuracy. The amount of data produced by an extended housekeeping system proves increasingly significant. However, to customers, housekeeping data is not of direct value and has therefore been subordinated to scientific payload data in terms of the allocation of bandwidth towards ground. In order to optimize the information throughput for a given bandwidth budget, data compression such as entropy coding as well as lossy data compaction need to be applied. At the same time, the accuracy and the allowed magnitude of error of housekeeping data is crucial to its value for ground engineers. As a result, especially lossy data compaction has to be applied carefully taking into account the nature of the data to be processed. In this paper, we evaluate transform-based compression techniques and analyze their effect on housekeeping data and suitability for subsequent entropy coding on board spacecrafts. To do so, we apply a variety of transforms to real sensor data collected by launchers (ARIANE5) as well as satellites (AISat) and analyze their performance in terms of data quality, compression ratio, computing effciency and effectiveness of subsequent entropy coding. Our results show that a data reduction of 96.5% for quickly oscilatting vibration sensors and of 99.5% for slower temperature sensors can be achieved without introducing a significant error during critical time frames within data sequences.

Current transients caused by energetic particle strikes are a serious threat for digital circuits in aerospace applications. Such single-event transients (SETs) can corrupt the circuit state, with possibly devastating consequences. Although it is possible to protect circuits with spatial redundancy techniques, the area and power overhead is high. Therefore aerospace circuits would benefit from adopting temporal redundancy instead, but existing solutions prioritize performance over reliability. Our proposed temporal redundancy latch-based architecture (TRLA) is a standard cell, static CMOS temporal redundancy technique, with area savings of 26%, power savings of 46%, and 14% faster circuit operation compared to triple modular redundancy (TMR).

High compression ratio is crucial to cope with the large amounts of data produced by telemetry sensors and the limited transmission bandwidth typical of space applications. A new generation of telemetry units is under development, based on Commercial Off-The-Shelf (COTS) components that may be subject to misbehaviors due to radiation-induced soft errors. The purpose of this paper is to study the impact of soft errors on different configurations of a discrete cosine transform (DCT)based compression algorithm. This work's main contribution lies in providing some design guidelines.

To gain deeper insight into the behavior, the mechanical and thermal stress and the environment of a spacecraft, extended monitoring is required. This implies the application of high resolution sensors in both value and time domain which produces a considerable amount of data.Thus, in order to optimize the information throughput of the bandwidth available for monitoring, we propose a compression algorithm derived from well-established approaches for image compression. Namely, a two-dimensional discrete cosine transform (DCT) will be applied to compress one-dimensional timeseries data of various sensor types.Data quality in terms of mean square error and peak signalto-noise ratio as well as compression ratio of the introduced algorithm together with its derived pareto optimal set of parameters will be verified by applying it to real-world sets of housekeeping data. We will use different sensor samples of both satellites (AISat) and launchers (ARIANE 5).

The increasing parallelism of many-core systems demands for efficient strategies for the run-time system management. Due to the large number of cores the management overhead has a rising impact to the overall system performance. This work analyzes a clustered infrastructure of dedicated hardware nodes to manage a homogeneous many-core system. The hardware nodes implement a message passing protocol and perform the task mapping and synchronization at run-time. To make meaningful mapping decisions, the global management nodes employ a workload status communication mechanism. This paper discusses the design-space of the dedicated infrastructure by means of task mapping use-cases and a parallel benchmark including application-interference. We evaluate the architecture in terms of application speedup and analyze the mechanism for the status communication. A comparison versus centralized and fully-distributed configurations demonstrates the reduction of the computation and communication management overhead for our approach.