Research Projects

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration

The aim of this project is to provide a formal logic- and model-based approach to support the development of dynamic control mechanisms that optimize the trade-off between resilience and its cost according to application-specific requirements. Our techniques will rely on logical reasoners that support the diagnosis of failures and run-time adaptions as well as probabilistic model checking techniques for the evaluation of resilience mechanisms under quantitative aspects. This requires appropriate modeling approaches that cover the relevant aspects of the system and their dominating cost factors, possible failures and the effect and cost of resilience policies. One major challenge for the application of logical reasoners is a proper treatment of scenarios, when only incomplete information is available (e.g., due to vague or unreliable sensor information). We will build three ontologies to represent (i) faults, (ii) resilience mechanisms, and (iii) user requirements, and connect them through an appropriate ontology linking approach. The goal is then to find a minimal sequence of actions that will maximize the information on the system, and in particular on the faulty components, building upon the ideas from abduction and model-based diagnosis. This knowledge will be used to deduce the best resilience mechanism to apply for the current setting.
Probabilistic model checking techniques will provide formal criteria for the evaluation and comparison of dynamic control mechanisms. These techniques should be compatible with the dynamic skeleton-approach developed by the the partners of orchestration and resilience path and should fit with the model checking approach used in the orchestration path. An additional challenge for the modeling approach and the application of model checking techniques are resilience-specific cost functions and optimization criteria. A particular challenge will be an adequate treatment of error propagation. First, in the cross-layer approach errors might propagate from one to the next layers. This type of error propagation might be handled using appropriate abstraction techniques that provide high-level representations of the neighboring layers and permit to treat them as environment constraints. Second, error propagation might be vertical between components. For the construction of models, this is in particular challenging if errors of individual components are not independent (e.g., common-cause errors). A novel and promising idea is to integrate the abduction- and model-based techniques for the failure diagnoses as basis for the generation of counterexamples that a probabilistic model checker might return and that might support the redesign or improvement of resilience strategies. In this context, we will also study the combination of fault injection and probabilistic model checking to derive assertions on the achieved degree of fault-tolerance for given control mechanisms.
The project will be divided in two stages that will run sequentially. In the first stage, we will focus on setting the formal logical basis for designing and deducing the best resilience policy to apply at each specific situation. At the second stage, we will apply model checking techniques to the evaluation and comparison of the different control mechanisms.

01/2013

Drittmittel

siehe Webseite

siehe Webseite

keine

http://

Nein

Nein

Nein

Failure Diagnosis, Reconfiguration