This thesis introduces an alternative Fault-Tolerant Power-Aware Hybrid Adder (or simply FARHAD) for high-performance processors. FARHAD, similar to earlier studies, relies on performing add operations twice to detect errors. Unlike previous studies, FARHAD uses an aggressive adder to produce the initial outcome and a low-power adder to generate the second outcome, referred to as the checker. FARHAD uses checkpoints, a feature already available to high-performance processors, to recover from errors. FARHAD achieves the high energy-efficiency of time-redundant solutions and the high performance of resource-redundant adders. We evaluate FARHAD from power and performance points of view using a subset of SPEC’2K benchmarks. Our evaluations show that FARHAD outperforms an alternative time-redundant solution by 20%. FARHAD reduces the power dissipation of an alternative resource-redundant adder by 40% while maintaining performance. / Graduate / 0544

The interaction of radiation with integrated circuits can provoke transient faults due to the deposit of charge in sensitive nodes of transistors. Because of the decrease the size in the process technology, charge sharing between transistors placed close to each other has been more and more observed. This phenomenon can lead to multiple transient faults. Therefore, it is important to analyze the effect of multiple transient faults in integrated circuits and investigate mitigation techniques able to cope with multiple faults. This work investigates the effect known as single-event-induced charge sharing in integrated circuits. Two main techniques are analyzed to cope with this effect. First, a placement constraint methodology is proposed. This technique uses placement constraints in standard cell based circuits. The objective is to achieve a layout for which the Soft-Error Rate (SER) due charge shared at adjacent cell is reduced. A set of fault injection was performed and the results show that the SER can be minimized due to single-event-induced charge sharing in according to the layout structure. Results show that by using placement constraint, it is possible to reduce the error rate from 12.85% to 10.63% due double faults. Second, Triple Modular Redundancy (TMR) schemes with different levels of granularities limited by majority voters are analyzed under multiple faults. The TMR versions are implemented using a standard design flow based on a traditional commercial standard cell library. An extensive fault injection campaign is then performed in order to verify the softerror rate due to single-event-induced charge sharing in multiple nodes. Results show that the proposed methodology becomes crucial to find the best trade-off in area, performance and soft-error rate when TMR designs are considered under multiple upsets. Results have been evaluated in a case-study circuit Advanced Encryption Standard (AES), synthesized to 90nm Application Specific Integrated Circuit (ASIC) library, and they show that combining the two techniques, the error rate resulted from multiple faults can be minimized or masked. By using TMR with different granularities and placement constraint methodology, it is possible to reduce the error rate from 11.06% to 0.00% for double faults. A detailed study of triple, four and five multiple faults combining both techniques are also described. We also tested the TMR with different granularities in SRAM-based FPGA platform. Results show that the versions with a fine grain scheme (FGTMR) were more effectiveness in masking multiple faults, similarly to results observed in the ASICs. In summary, the main contribution of this master thesis is the investigation of charge sharing effects in ASICs and the use of a combination of techniques based on TMR redundancy and placement to improve the tolerance under multiple faults.

The interaction of radiation with integrated circuits can provoke transient faults due to the deposit of charge in sensitive nodes of transistors. Because of the decrease the size in the process technology, charge sharing between transistors placed close to each other has been more and more observed. This phenomenon can lead to multiple transient faults. Therefore, it is important to analyze the effect of multiple transient faults in integrated circuits and investigate mitigation techniques able to cope with multiple faults. This work investigates the effect known as single-event-induced charge sharing in integrated circuits. Two main techniques are analyzed to cope with this effect. First, a placement constraint methodology is proposed. This technique uses placement constraints in standard cell based circuits. The objective is to achieve a layout for which the Soft-Error Rate (SER) due charge shared at adjacent cell is reduced. A set of fault injection was performed and the results show that the SER can be minimized due to single-event-induced charge sharing in according to the layout structure. Results show that by using placement constraint, it is possible to reduce the error rate from 12.85% to 10.63% due double faults. Second, Triple Modular Redundancy (TMR) schemes with different levels of granularities limited by majority voters are analyzed under multiple faults. The TMR versions are implemented using a standard design flow based on a traditional commercial standard cell library. An extensive fault injection campaign is then performed in order to verify the softerror rate due to single-event-induced charge sharing in multiple nodes. Results show that the proposed methodology becomes crucial to find the best trade-off in area, performance and soft-error rate when TMR designs are considered under multiple upsets. Results have been evaluated in a case-study circuit Advanced Encryption Standard (AES), synthesized to 90nm Application Specific Integrated Circuit (ASIC) library, and they show that combining the two techniques, the error rate resulted from multiple faults can be minimized or masked. By using TMR with different granularities and placement constraint methodology, it is possible to reduce the error rate from 11.06% to 0.00% for double faults. A detailed study of triple, four and five multiple faults combining both techniques are also described. We also tested the TMR with different granularities in SRAM-based FPGA platform. Results show that the versions with a fine grain scheme (FGTMR) were more effectiveness in masking multiple faults, similarly to results observed in the ASICs. In summary, the main contribution of this master thesis is the investigation of charge sharing effects in ASICs and the use of a combination of techniques based on TMR redundancy and placement to improve the tolerance under multiple faults.

The interaction of radiation with integrated circuits can provoke transient faults due to the deposit of charge in sensitive nodes of transistors. Because of the decrease the size in the process technology, charge sharing between transistors placed close to each other has been more and more observed. This phenomenon can lead to multiple transient faults. Therefore, it is important to analyze the effect of multiple transient faults in integrated circuits and investigate mitigation techniques able to cope with multiple faults. This work investigates the effect known as single-event-induced charge sharing in integrated circuits. Two main techniques are analyzed to cope with this effect. First, a placement constraint methodology is proposed. This technique uses placement constraints in standard cell based circuits. The objective is to achieve a layout for which the Soft-Error Rate (SER) due charge shared at adjacent cell is reduced. A set of fault injection was performed and the results show that the SER can be minimized due to single-event-induced charge sharing in according to the layout structure. Results show that by using placement constraint, it is possible to reduce the error rate from 12.85% to 10.63% due double faults. Second, Triple Modular Redundancy (TMR) schemes with different levels of granularities limited by majority voters are analyzed under multiple faults. The TMR versions are implemented using a standard design flow based on a traditional commercial standard cell library. An extensive fault injection campaign is then performed in order to verify the softerror rate due to single-event-induced charge sharing in multiple nodes. Results show that the proposed methodology becomes crucial to find the best trade-off in area, performance and soft-error rate when TMR designs are considered under multiple upsets. Results have been evaluated in a case-study circuit Advanced Encryption Standard (AES), synthesized to 90nm Application Specific Integrated Circuit (ASIC) library, and they show that combining the two techniques, the error rate resulted from multiple faults can be minimized or masked. By using TMR with different granularities and placement constraint methodology, it is possible to reduce the error rate from 11.06% to 0.00% for double faults. A detailed study of triple, four and five multiple faults combining both techniques are also described. We also tested the TMR with different granularities in SRAM-based FPGA platform. Results show that the versions with a fine grain scheme (FGTMR) were more effectiveness in masking multiple faults, similarly to results observed in the ASICs. In summary, the main contribution of this master thesis is the investigation of charge sharing effects in ASICs and the use of a combination of techniques based on TMR redundancy and placement to improve the tolerance under multiple faults.

Partial circuit replication is a soft error mitigation technique that uses redundant copies of a circuit to mask or detect the effects of soft errors. By masking or detecting the effect of soft errors on SRAM-based FPGAs, implemented circuits can be made more reliable. The technique is applied selectively, to only a portion of the components within a circuit. Partial application lowers the cost of implementation. The objective of partial circuit replication is to provide maximal benefit at limited or minimized cost. The greatest challenge of partial circuit replication is selecting which components within a circuit to replicate. This dissertation advances the state of the art in the effective use of partial circuit replication for masking and detecting soft errors in SRAM-based FPGAs. It provides a theoretical foundation in which the expected benefits and challenges of partial circuit replication can be understood. It proposes several new selection approaches for identifying the most beneficial areas of a circuit to replicate. These approaches are applied to two complex FPGA-based computer networking systems and another FPGA design. The effectiveness of the selection approaches are evaluated through fault injection and accelerated radiation testing. More benefit than expected is obtained through partial circuit replication when applied to critical components and sub-regions of the designs. In one example, in an open-source computer networking design, partial circuit replication masks and detects approximately 70% of failures while replicating only 5% of circuit components, a benefit-cost ratio of 14.0.

Underwater navigation faces many problems with accurately estimating the absolute positionof an underwater vehicle. Neither Global Positioning system (GPS) nor Long Baseline (LBL) orShort Baseline (SBL) are possible to use for a military vehicle acting under stealth, since thesetechniques require the vehicle to be in the vicinity of a nearby ship or to surface and raise its antenna. It will therefore have to rely on sensors such as Doppler Velocity Log (DVL) and a compassto estimate its absolute position using dead reckoning or an Inertial Navigation System (INS). Thisthesis presents an alternative Multiple model Kalman Filter (KF) to the existing Multiple ModelAdaptive Estimator (MMAE) algorithm using n-Modular Redundancy (NMR), in order to gaina more accurate result than with a single KF. By analyzing how different amounts of filters andvoter types affect the accuracy and precision of the velocity and heading estimations, the potentialbenefits and drawbacks can be drawn for each solution. Such benefits and drawbacks were alsovisually evaluated in a Matlab script which was used to calculate the coordinates using the velocityand heading from the speed sensors and compass, without the need for running the filtered states onthe vehicle’s navigation system. The results present the potential of using a multiple model KF inthe form of an NMR, which was demonstrated by both the amount of reduced noise in the velocitystates and how the filters were used in a virtual navigation system created in Matlab.

Triple Modular Redundancy (TMR) is a common reliability technique for mitigating single event upsets (SEUs) in FPGA designs operating in radiation environments. For FPGA systems that employ configuration scrubbing, majority voters are needed in all feedback paths to ensure proper synchronization between the TMR replicates. Synchronization voters, however, consume additional resources and impact system timing. This work introduces and contrasts seven algorithms for inserting synchronization voters while automatically performing TMR. The area cost and timing impact of each algorithm on a number of circuit benchmarks is reported. The work demonstrates that one of the algorithms provides the best overall timing performance results with an average 8.5% increase in critical path length over a triplicated design without voters and a 29.6% area increase. Another algorithm provides far better area results (an average 3.4% area increase over a triplicated design without voters) at a slightly higher timing cost (an average 14.9% increase in critical path length over a triplicated design without voters). In addition, this work demonstrates that restricting synchronization voter locations to flip-flop output nets is an effective heuristic for minimizing the timing performance impact of synchronization voter insertion.

Semiconductor manufacturing defects adversely affect yield and reliability. Manufacturers expend vast resources to reduce defects within their processes. As the minimum feature size get smaller, defects become increasingly difficult to prevent. Defects can change the behavior of a logic circuit resulting in a fault. Manufacturers and designers may improve yield, reliability, and profitability by using design techniques that make products robust even in the presence of faults. Triple modular redundancy (TMR) is a fault tolerant technique commonly used to mask faults using voting outcomes from three processing elements (PE). TMR is effective at masking errors as long as no more than a single processing element is faulty. Time distributed voting (TDV) is proposed as an active fault tolerant technique. TDV addresses the shortcomings of triple modular redundancy (TMR) in the presence of multiple faulty processing elements. A faulty PE may not be incorrect 100% of the time. When a faulty element generates correct results, a majority is formed with the healthy PE. TDV observes voting outcomes over time to make a statistical decision whether a PE is healthy or faulty. In simulation, fault coverage is extended to 98.6% of multiple faulty PE cases. As an active fault tolerant technique, TDV identifies faulty PE's so that actions may be taken to replace or disable them in the system. TDV may provide a positive impact to semiconductor manufacturers by improving yield and reliability even as fault frequency increases.

Radiation effects encountered in space or aviation environments can affect the configuration bits in Field Programmable Gate Arrays (FPGA) causing errors in FPGA output. One method of increasing FPGA reliability in radiation environments includes adding redundant logic to mask errors and allow time for repair. Despite the redundancy added with triple modular redundancy (TMR) and configuration scrubbing there exist some configuration bits that individually affect multiple TMR domains causing errors in FPGA output. A new tool called DeBit is introduced that identifies hardware resources associated with a single bit failure. This tool identifies a novel failure mode involving global routing resources and the failure mode is verified through a series of directed tests on global routing resources. Lastly, a mitigation strategy is proposed and tested on a single error in a triple modular redundancy (TMR) design.

Radiation effects encountered in space or aviation environments can affect the configuration bits in Field Programmable Gate Arrays (FPGA) causing errors in FPGA output. One method of increasing FPGA reliability in radiation environments includes adding redundant logic to mask errors and allow time for repair. Despite the redundancy added with triple modular redundancy (TMR) and configuration scrubbing there exist some configuration bits that individually affect multiple TMR domains causing errors in FPGA output. A new tool called DeBit is introduced that identifies hardware resources associated with a single bit failure. This tool identifies a novel failure mode involving global routing resources and the failure mode is verified through a series of directed tests on global routing resources. Lastly, a mitigation strategy is proposed and tested on a single error in a triple modular redundancy (TMR) design.