A framework for fault tolerant real time systems based on reconfigurable FPGAs

To increase the amount of logic available to the users in SRAM-based FPGAs, manufacturers are using nanometric technologies to boost logic density and reduce costs, making its use more attractive. However, these technological improvements also make FPGAs particularly vulnerable to configuration memory bit-flips caused by power fluctuations, strong electromagnetic fields and radiation. This issue is particularly sensitive because of the increasing amount of configuration memory cells needed to define their functionality. A short survey of the most recent publications is presented to support the options assumed during the definition of a framework for implementing circuits immune to bit-flips induction mechanisms in memory cells, based on a customized redundant infrastructure and on a detection-and-fix controller.

Multi-μ: an Ada 95 based architecture for fault tolerance support of real-time systems

This paper presents an architecture (Multi-μ) being implemented to study and develop software based fault tolerant mechanisms for Real-Time Systems, using the Ada language (Ada 95) and Commercial Off-The-Shelf (COTS) components. Several issues regarding fault tolerance are presented and mechanisms to achieve fault tolerance by software active replication in Ada 95 are discussed. The Multi-μ architecture, based on a specifically proposed Fault Tolerance Manager (FTManager), is then described. Finally, some considerations are made about the work being done and essential future developments.

A comparative analysis of fault injection methods via enhanced on-chip debug infrastructures

On-chip debug (OCD) features are frequently available in modern microprocessors. Their contribution to shorten the time-to-market justifies the industry investment in this area, where a number of competing or complementary proposals are available or under development, e.g. NEXUS, CJTAG, IJTAG. The controllability and observability features provided by OCD infrastructures provide a valuable toolbox that can be used well beyond the debugging arena, improving the return on investment rate by diluting its cost across a wider spectrum of application areas. This paper discusses the use of OCD features for validating fault tolerant architectures, and in particular the efficiency of various fault injection methods provided by enhanced OCD infrastructures. The reference data for our comparative study was captured on a workbench comprising the 32-bit Freescale MPC-565 microprocessor, an iSYSTEM IC3000 debugger (iTracePro version) and the Winidea 2005 debugging package. All enhanced OCD infrastructures were implemented in VHDL and the results were obtained by simulation within the same fault injection environment. The focus of this paper is on the comparative analysis of the experimental results obtained for various OCD configurations and debugging scenarios.

Real time fault injection using a modified debugging infrastructure

Dependability is a critical factor in computer systems, requiring high quality validation & verification procedures in the development stage. At the same time, digital devices are getting smaller and access to their internal signals and registers is increasingly complex, requiring innovative debugging methodologies. To address this issue, most recent microprocessors include an on-chip debug (OCD) infrastructure to facilitate common debugging operations. This paper proposes an enhanced OCD infrastructure with the objective of supporting the verification of fault-tolerant mechanisms through fault injection campaigns. This upgraded on-chip debug and fault injection (OCD-FI) infrastructure provides an efficient fault injection mechanism with improved capabilities and dynamic behavior. Preliminary results show that this solution provides flexibility in terms of fault triggering and allows high speed real-time fault injection in memory elements

A framework for self-healing radiation-tolerant implementations on reconfigurable FPGAs

To increase the amount of logic available in SRAM-based FPGAs manufacturers are using nanometric technologies to boost logic density and reduce prices. However, nanometric scales are highly vulnerable to radiation-induced faults that affect values stored in memory cells. Since the functional definition of FPGAs relies on memory cells, they become highly prone to this type of faults. Fault tolerant implementations, based on triple modular redundancy (TMR) infrastructures, help to keep the correct operation of the circuit. However, TMR is not sufficient to guarantee the safe operation of a circuit. Other issues like the effects of multi-bit upsets (MBU) or fault accumulation, have also to be addressed. Furthermore, in case of a fault occurrence the correct operation of the affected module must be restored and the current state of the circuit coherently re-established. A solution that enables the autonomous correct restoration of the functional definition of the affected module, avoiding fault accumulation, re-establishing the correct circuit state in realtime, while keeping the normal operation of the circuit, is presented in this paper.

A modified debugging infrastructure to assist real time fault injection campaigns

Fault injection is frequently used for the verification and validation of the fault tolerant features of microprocessors. This paper proposes the modification of a common on-chip debugging (OCD) infrastructure to add fault injection capabilities and improve performance. The proposed solution imposes a very low logic overhead and provides a flexible and efficient mechanism for the execution of fault injection campaigns, being applicable to different target system architectures.

On-line self-healing of circuits implemented on reconfigurable FPGAs

To boost logic density and reduce per unit power consumption SRAM-based FPGAs manufacturers adopted nanometric technologies. However, this technology is highly vulnerable to radiation-induced faults, which affect values stored in memory cells, and to manufacturing imperfections. Fault tolerant implementations, based on Triple Modular Redundancy (TMR) infrastructures, help to keep the correct operation of the circuit. However, TMR is not sufficient to guarantee the safe operation of a circuit. Other issues like module placement, the effects of multi- bit upsets (MBU) or fault accumulation, have also to be addressed. In case of a fault occurrence the correct operation of the affected module must be restored and/or the current state of the circuit coherently re-established. A solution that enables the autonomous restoration of the functional definition of the affected module, avoiding fault accumulation, re-establishing the correct circuit state in real-time, while keeping the normal operation of the circuit, is presented in this paper.