Aldeida Aleti

Software Architecture Optimization Methods: A Systematic Literature Review

Due to significant industrial demands toward software systems with increasing complexity and challenging quality requirements, software architecture design has become an important development activity and the research domain is rapidly evolving. In the last decades, software architecture optimization methods, which aim to automate the search for an optimal architecture design with respect to a (set of) quality attribute(s), have proliferated. However, the reported results are fragmented over different research communities, multiple system domains, and multiple quality attributes. To integrate the existing research results, we have performed a systematic literature review and analyzed the results of 188 research papers from the different research communities. Based on this survey, a taxonomy has been created which is used to classify the existing research. Furthermore, the systematic analysis of the research literature provided in this review aims to help the research community in consolidating the existing research efforts and deriving a research agenda for future developments.

ArcheOpterix: An extendable tool for architecture optimization of AADL models

For embedded systems quality requirements are equally if not even more important than functional requirements. The foundation for the fulfillment of these quality requirements has to be set in the architecture design phase. However, finding a suitable architecture design is a difficult task for software and system architects. Some of the reasons for this are an ever-increasing complexity of todays systems, strict design constraints and conflicting quality requirements. To simplify the task, this paper presents an extendable Eclipse-based tool, called ArcheOpterix, which provides a framework to implement evaluation techniques and optimization heuristics for AADL specifications. Currently, evolutionary strategies have been implemented to identify optimized deployment architectures with respect to multiple quality objectives and design constraints. Experiments with a set of initial deployment architectures provide evidence that the tool can successfully find architecture specifications with better quality.

Reliability-driven deployment optimization for embedded systems

One of the crucial aspects that influence reliability of embedded systems is the deployment of software components to hardware nodes. If the hardware architecture is designed prior to the customized software architecture, which is often the case in product-line manufacturing (e.g. in the automotive domain), the system architect needs to resolve a nontrivial task of finding a (near-)optimal deployment balancing the reliabilities of individual services implemented on the software level.In this paper, we introduce an approach to automate this task. As distinct to related approaches, which typically stay with quantification of reliability for a specific deployment, we target multi-criteria optimization and provide the architect with near-optimal (non-dominated) deployment alternatives with respect to service reliabilities. Toward this goal, we annotate the software and hardware architecture with necessary reliability-relevant attributes, design a method to quantify the quality of individual deployment alternatives, and implement the approach employing an evolutionary algorithm.

Architecture-Driven reliability and energy optimization for complex embedded systems

The use of redundant computational nodes is a widely used design tactic to improve the reliability of complex embedded systems. However, this redundancy allocation has also an effect on other quality attributes, including energy consumption, as each of the redundant computational nodes requires additional energy. As a result, the two quality objectives are conflicting. The approach presented in this paper applies a multi-objective optimization strategy to find optimal redundancy levels for different architectural elements. It is implemented in the ArcheOpterix tool and illustrated on a realistic case study from the automotive domain.

Architecture-based reliability evaluation under uncertainty

The accuracy of architecture-based reliability evaluations depends on a number of parameters that need to be estimated, such as environmental factors or system usage. Researchers have tackled this problem by including uncertainties in architecture evaluation models and solving them analytically and with simulations. The usual assumption is that the input parameter distributions are normal, and that it is sufficient to report the attributes that describe the system in terms of the mean and variance of the attribute. In this work, we introduce a simulation-based approach that can accommodate a diverse set of parameter range distributions, self-regulate the number of architecture evaluations to the desired significance level and reports the desired percentiles of the values which ultimately characterise a specific quality attribute of the system. We include a case study which illustrates the flexibility of this approach using the evaluation of system reliability.

Let the Ants Deploy Your Software - An ACO Based Deployment Optimisation Strategy

Decisions regarding the mapping of software components to hardware nodes affect the quality of the resulting system. Making these decisions is hard when considering the ever-growing complexity of the search space, as well as conflicting objectives and constraints. An automation of the solution space exploration would help not only to make better decisions but also to reduce the time of this process. In this paper, we propose to employ Ant Colony Optmisation (ACO) as a multi-objective optimisation strategy. The constructive approach is compared to an iterative optimisation procedure - a Genetic Algorithm (GA) adaptation - and was observed to perform suprisingly similar, although not quite on a par with the GA, when validated based on a series of experiments.

Model-based performance analysis of software architectures under uncertainty

Performance analysis is often conducted before achieving full knowledge of a software system, in other words under a certain degree of uncertainty. Uncertainty is particularly critical in the performance domain when it relates to values of parameters such as workload, operational profile, resource demand of services, service time of hardware devices, etc. The goal of this paper is to explicitly consider uncertainty in the performance modelling and analysis process. In particular, we use probabilistic formulation of parameter uncertainties and present a Monte Carlo simulation-based approach to systematically assess the robustness of an architectural model despite its uncertainty. In case of unsatisfactory results, we introduce refactoring actions aimed at generating new software architectural models that better tolerate the uncertainty of parameters. The proposed approach is illustrated on a case study from the e-Health domain.

Architecture-driven reliability optimization with uncertain model parameters

Highlights? An architecture optimization approach that considers uncertainty in input parameters. ? Search for good solutions that restrict the impact of parameter uncertainties. ? Tested the new approach on a case study of component deployment problem. ? Results of a series of experiments on performance and scalability of the approach. It is currently considered good software engineering practice to decide between design alternatives based on quantitative architecture evaluations for different quality attributes, such as reliability and performance. However, the results of these quantitative architecture evaluations are dependent on design-time estimates for a series of model-parameters, which may not be accurate and have to be estimated subject heterogeneous uncertain factors. As a result, sub-optimal design decisions may be taken. To overcome this problem, we present a novel robust optimization approach that deals with parameter uncertainties at the design phase of software-intensive systems. This work specifically focuses on architecture-based reliability evaluation models. The proposed approach is able to find good solutions that restrict the impact of parameter uncertainties, and thus provides better decision support. The accuracy and scalability of the presented approach is validated with an industrial case study and a series of experiments with generated examples in different problem sizes.

Robust ArcheOpterix: architecture optimization of embedded systems under uncertainty

Design of embedded systems involves a number of architecture decisions which have a significant impact on its quality. Due to the complexity of todays systems and the large design options that need to be considered, making these decisions is beyond the capabilities of human comprehension and makes the architectural design a challenging task. Several tools and frameworks have been developed, which automate the search for optimal or near-optimal design decisions based on quantitative architecture evaluations for different quality attributes. However, current approaches use approximations for a series of model parameters which may not be accurate and have to be estimated subject to heterogeneous uncertain factors. We have developed a framework which considers the uncertainty of design-time parameter estimates, and optimizes embedded system architectures for robust quality goals. The framework empowers conventional architecture optimization approaches with modeling and tool support for architecture description, model evaluation and architecture optimization on the face of uncertainty.

Test data generation with a Kalman filter-based adaptive genetic algorithm

We introduce a new approach for generating test data, based on adaptive optimisation.The adaptive optimisation framework uses feedback from the optimisation process to adjust parameter values of a genetic algorithm during the search.Our approach is compared to a state of the art test data optimisation algorithm that does not adapt parameter values online, and a conspicuous adaptive optimisation algorithm, outperforming both methods in a wide range of problems. Software testing is a crucial part of software development. It enables quality assurance, such as correctness, completeness and high reliability of the software systems. Current state-of-the-art software testing techniques employ search-based optimisation methods, such as genetic algorithms to handle the difficult and laborious task of test data generation. Despite their general applicability, genetic algorithms have to be parameterised in order to produce results of high quality. Different parameter values may be optimal for different problems and even different problem instances. In this work, we introduce a new approach for generating test data, based on adaptive optimisation. The adaptive optimisation framework uses feedback from the optimisation process to adjust parameter values of a genetic algorithm during the search. Our approach is compared to a state of the art test data optimisation algorithm that does not adapt parameter values online, and a representative adaptive optimisation algorithm, outperforming both methods in a wide range of problems.