Network Service Providers are struggling to reduce cost and still improve customer satisfaction. We have looked at three underlying challenges to achieve these goals; an overwhelming flow of low-quality alarms, understanding the structure and quality of the delivered services, and automation of service configuration. This thesis proposes solutions in these areas based on domain-specific languages, data-mining and self-learning. Most of the solutions have been validated based on data from a large service provider. We look at how domain-models can be used to capture explicit knowledge for alarms and services. In addition, we apply data-mining and self-learning techniques to capture tacit knowledge. The validation shows that models improve the quality of alarm and service models, and enables automatic rendering of functions like root cause correlation, service and SLA status, as well as service configuration. The data-mining and self-learning solutions show that we can learn from available decisions made by experts and automatically assign alarm priorities.

Supercompilation algorithms can perform great optimizations but sometimes suffer from the problem of code explosion. This results in huge binaries which might hurt the performance on a modern processor. We present a supercompilation algorithm that is fast enough to speculatively supercompile expressions and discard the result if it turned out bad. This allows us to supercompile large parts of the imaginary and spectral parts of nofib in a matter of seconds while keeping the binary size increase below 5%.

Previous deforestation and supercompilation algorithms may introduce accidental termination when applied to call-by-value programs. This hides looping bugs from the programmer, and changes the behavior of a program depending on whether it is optimized or not. We present a supercompilation algorithm for a higher-order call-by-value language and we prove that the algorithm both terminates and preserves termination properties. This algorithm utilizes strictness information for deciding whether to substitute or not and compares favorably with previous call-by-name transformations.

Embedded systems are often operating under hard real-time constraints. Such systems are naturally described as time-bound reactions to external events, a point of view made manifest in the high-level programming and systems modeling language Timber. In this paper we demonstrate how the Timber semantics for parallel reactive objects translates to embedded real-time programming in C. This is accomplished through the use of a minimalistic Timber Run-Time system, TinyTimber (TT). The TT kernel ensures state integrity, and performs scheduling of events based on given time-bounds in compliance with the Timber semantics. In this way, we avoid the volatile task of explicitly coding parallelism in terms of processes/threads/semaphores/monitors, and side-step the delicate task to encode time-bounds into priorities. In this paper, the TT kernel design is presented and performance metrics are presented for a number of representative embedded platforms, ranging from small 8-bit to more potent 32- bit micro controllers. The resulting system runs on bare metal, completely free of references to external code (even C-lib) which provides a solid basis for further analysis. In comparison to a traditional thread based real-time operating system for embedded applications (FreeRTOS), TT has tighter timing performance and considerably lower code complexity. In conclusion, TinyTimber is a viable alternative for implementing embedded real-time applications in C today.

Designing a garbage collector with real-time properties is a particularlydifficult task, involving the construction of both an incremental run-timealgorithm as well as methods enabling a priori reasoning about schedulability in two dimensions (time and memory usage in conjunction). In order to comply with such ambitious goals with any amount of formal rigor, a comprehensive understanding of the actual algorithm used is of course a fundamental requirement. In this paper we present a formal model of an incremental copying garbage collector, where each atomic increment is modeled as a transition between states of a heap process. Soundness of the algorithm is shown by proving that the garbage collecting heap process is weakly bisimilar to a non-collecting heap with infinite storage space. In addition, we show that our collector is both terminating and useful, in the sense that it actually recovers the unreachable parts of any given heap in a finite number of steps.

We present a subtyping extension to the Hindley/Milner type system that is based on name inequivalence. This approach allows the subtype relation to be defined by incremental construction of polymorphic records and datatypes, in a way that subsumes the basic type systems of both languages like ML and Java. As the main contribution of the paper, we describe a partial type inference algorithm for the extended system which favours succinctness over generality, in the sense that it never infers types with subtype constraints. The algorithm is based on an efficient approximating constraint solver, and is able to type a wide range of programs that utilize subtyping and polymorphism in a non-trivial way. Since constrained types are not inferred, the algorithm cannot be complete; however, we provide a completeness result w.r.t. the Hindley/Milner type system as a form of characterizing lower bound.