<p>A <i>method</i> for designing workflow systems which satisfy performance requirements is proposed in this thesis. Integration of human and computerised performance is particularly useful for workflow systems where human and computerised processes are intertwined. The proposed <i>framework </i>encompasses human and computerised resources. </p><p>Even though systematic performance engineering is not common practice in information system development, current, <i>best</i> practice shows that performance engineering of software is feasible, e.g. the SPE method by Connie U. Smith. Contemporary approaches to performance engineering focus on <i>dynamic </i>models of resource contention, e.g. queueing networks and Petri nets. Two difficulties arise for large-scale information systems. The first difficulty is to estimate appropriate <i>parameters </i>which capture the properties of the software and the organisation. The second difficulty is to maintain an <i>overveiw </i>of a complex model, which is essential both to guide the choice of parameters and to ensure that the oerformance engineering process is an intregal part of the wider system development process.</p><p>The proposed method is based on the <i>static</i> performance modelling method Structure and Performance (SP) developed by Peter h: Hughes. SP provides a suitable bridge between contemporary CASE tools and traditional dynamic approaches to performance evaluation, in particular because it adresses the problems of parameterisation and overveiw identified above.</p><p>The method is explored and illustrated with two case studies. The <i>Blood Bank Case Study</i> comprised performance engineering of a transaction-oriented information system, showing the oractical feasibility of integrating the method with CASE tools. The <i>Gas Sales</i> <i>Telex Administration Case Study</i> for Statoil looked at performance engineering of a workfloww system for telex handling, and consisted of performance modelling of human activity in interaction with a Lotus Notes computer platform.</p><p>The latter case study demonstrated the feasibility of the framework.</p>

A method for designing workflow systems which satisfy performance requirements is proposed in this thesis. Integration of human and computerised performance is particularly useful for workflow systems where human and computerised processes are intertwined. The proposed framework encompasses human and computerised resources. Even though systematic performance engineering is not common practice in information system development, current, best practice shows that performance engineering of software is feasible, e.g. the SPE method by Connie U. Smith. Contemporary approaches to performance engineering focus on dynamic models of resource contention, e.g. queueing networks and Petri nets. Two difficulties arise for large-scale information systems. The first difficulty is to estimate appropriate parameters which capture the properties of the software and the organisation. The second difficulty is to maintain an overveiw of a complex model, which is essential both to guide the choice of parameters and to ensure that the oerformance engineering process is an intregal part of the wider system development process. The proposed method is based on the static performance modelling method Structure and Performance (SP) developed by Peter h: Hughes. SP provides a suitable bridge between contemporary CASE tools and traditional dynamic approaches to performance evaluation, in particular because it adresses the problems of parameterisation and overveiw identified above. The method is explored and illustrated with two case studies. The Blood Bank Case Study comprised performance engineering of a transaction-oriented information system, showing the oractical feasibility of integrating the method with CASE tools. The Gas Sales Telex Administration Case Study for Statoil looked at performance engineering of a workfloww system for telex handling, and consisted of performance modelling of human activity in interaction with a Lotus Notes computer platform. The latter case study demonstrated the feasibility of the framework.

<p>The ability to construct software, call it a functional ciphertext, which can be remotely executed in encrypted form as an entirely self-contained unit, has the potential for some interesting applications. One such application is the construction of autonomous mobile agents capable of entering into certain types of legally binding contracts on behalf of the sender. At a premium in such circumstances is the ability to protect secret cryptographic keys or other secret information, which typically is necessary for legally binding contracts. Also important is the ability to do powerful computations, that are more than just one-off secure function evaluations.</p><p>The problem of constructing computation systems that achieve this, has been attempted by many to little or no avail. This thesis presents three similar cryptographic systems that take a step closer to making such encrypted software a reality.</p><p>First is demonstrated how one can construct mappings from finite automata, that through iteration can do computations. A stateless storage construction, called a Turing platform, is defined and it is shown that such a platform, in conjunction with a functional representation of a finite automaton, can perform Turing universal computation. </p><p>The univariate, multivariate, and parametric ciphers for the encryption of multivariate mappings are presented and cryptanalyzed. Cryptanalysis of these ciphers shows that they must be used very carefully, in order to resist cryptanalysis. Entirely new to cryptography is the ability to remotely and securely re-encrypt functional ciphertexts made with either univariate or multivariate encryption.</p><p>Lastly it is shown how the ciphers presented can be applied to the automaton representations in the form of mappings, to do general encrypted computation. Note: many of the novel constructions in this thesis are covered by a patent application.</p>

The ability to construct software, call it a functional ciphertext, which can be remotely executed in encrypted form as an entirely self-contained unit, has the potential for some interesting applications. One such application is the construction of autonomous mobile agents capable of entering into certain types of legally binding contracts on behalf of the sender. At a premium in such circumstances is the ability to protect secret cryptographic keys or other secret information, which typically is necessary for legally binding contracts. Also important is the ability to do powerful computations, that are more than just one-off secure function evaluations. The problem of constructing computation systems that achieve this, has been attempted by many to little or no avail. This thesis presents three similar cryptographic systems that take a step closer to making such encrypted software a reality. First is demonstrated how one can construct mappings from finite automata, that through iteration can do computations. A stateless storage construction, called a Turing platform, is defined and it is shown that such a platform, in conjunction with a functional representation of a finite automaton, can perform Turing universal computation. The univariate, multivariate, and parametric ciphers for the encryption of multivariate mappings are presented and cryptanalyzed. Cryptanalysis of these ciphers shows that they must be used very carefully, in order to resist cryptanalysis. Entirely new to cryptography is the ability to remotely and securely re-encrypt functional ciphertexts made with either univariate or multivariate encryption. Lastly it is shown how the ciphers presented can be applied to the automaton representations in the form of mappings, to do general encrypted computation. Note: many of the novel constructions in this thesis are covered by a patent application.