Geir Horn

A development framework and methodology for self-adapting applications in ubiquitous computing environments

Today software is the main enabler of many of the appliances and devices omnipresent in our daily life and important for our well being and work satisfaction. It is expected that the software works as intended, and that the software always and everywhere provides us with the best possible utility. This paper discusses the motivation, technical approach, and innovative results of the MUSIC project. MUSIC provides a comprehensive software development framework for applications that operate in ubiquitous and dynamic computing environments and adapt to context changes. Context is understood as any information about the user needs and operating environment which vary dynamically and have an impact on design choices. MUSIC supports several adaptation mechanisms and offers a model-driven application development approach supported by a sophisticated middleware that facilitates the dynamic and automatic adaptation of applications and services based on a clear separation of business logic, context awareness and adaptation concerns. The main contribution of this paper is a holistic, coherent presentation of the motivation, design, implementation, and evaluation of the MUSIC development framework and methodology.

A component-based planning framework for adaptive systems

Recently, many researchers have focused on designing generic and reusable middlewares to overcome the complexity in building adaptive systems There is a general agreement that the openness provided by component-based approaches coupled with reflection mechanisms is the minimum prerequisites for supporting dynamic reconfigurations However, this is not sufficient to implement the heart of the adaptation loop namely the decision making on the required reconfiguration that adapts the system in a given context In this regard, this paper proposes a planning framework that subsumes and automates the adaptation decision-making in reflective component-based adaptive systems The salient feature of this framework is to model the variability of the adaptive system as a set of variation points at which alternative component compositions and implementations can be selected to form an application configuration The selection of a feasible configuration in a given context is based on the concept of component wise utility functions that estimates the user benefit of including a specific implementation alternative at a variation point in the composition We show that the selection problem can be modelled as a multi constraint shortest path that can be found in polynomial time Our approach is validated through a real world example implementing adaptive scenarios in the domain of mobile computing.

Software Agents for Collaborating Smart Solar-Powered Micro-Grids

Solar electricity is one of the options as innovative approach as primary energy use. It could be deployed decentralised into the urban areas, and could alleviate the carbonised electricity demand drastically. Information and communication technologies (ICT) could be exploited to provide real time information on energy consumption in a home or a building giving the possibility to citizens to take decisions in order to save energy. In this context CoSSMic, an ICT European project, aims at fostering a higher rate of self-consumption of decentralised renewable energy production by innovative autonomic systems for management and control of power micro-grids on users behalf. The paper addresses these challenges and discusses related work dealing with the development of an ICT solution using software agents which collaborate in a neighborhood, and with the central power grid, over a peer-to-peer overlay.

Ant-Based Approach to the Quality Aware Application Service Partitioning in a Grid Environment

This paper presents the first ant system based approach to the problem of configuring an application seen as a service composition on a distributed grid execution platform. The service set is to be partitioned onto the available execution nodes such that they satisfy some minimum quality requirements. Fundamentally this is an NP-hard problem. This paper explores the metaphor of foraging unintelligent ants of an ant colony, and model a de-centralised multi-agent method for solving the service partitioning problem. Moreover this approach establishes a distributed problem solving mechanism which does not require to have a central control. Simulations show that this is a viable approach for configuring applications in a grid environment.

A vision for better cloud applications

In this paper, we provide an overview over the PaaSage projects approach to helping the developer in exploiting cloud environments according to their specific needs and requirements. Classical software engineering methodologies no longer apply in multi-tenant, elastic environments, if the full capabilities for cost reduction and availability are to be exploited. PaaSage aims at offering software engineering extensions covering the full application lifecycle from deployment to execution.

A vision for a stochastic reasoner for autonomic cloud deployment

Applications deployed in multi-clouds will face issues like where to deploy the different artefacts, how to scale the application in case of performance problems, and how to adapt the application deployment. For complex applications it may be difficult to find manually the best allocation of the artefacts on the available infrastructures. This paper presents a vision for an autonomic deployment system. In particular, it details the architecture of a learning automata based reasoning component envisioned to be able to provide feasible allocations and discusses the research challenges originating from this approach.

A distributed agent-based system for coordinating smart solar-powered microgrids

Renewable energy like solar power is crucial for the transition to more sustainable energy supply and use in the modern society. Buildings with rooftop solar panels form microgrids acting as prosumers and are usually not under the control of the regulated companies operating the public grids. Currently much work has focused on the self-consumption of individual microgrids. On the contrary, the CoSSMic system targets at a neighborhood of microgrids with the primary goal to maximize the self-consumption of the whole neighborhood by co-ordinating their energy use and storage. To address the challenge of the fluctuating and partly unpredictable nature of renewable energy, a novel hybrid control mechanism is proposed, where planning and scheduling based on predictions is supplemented by a reactive feedback loop to compensate the inability to predict accurately the rapid fluctuations in PV output due to passing clouds. To enable easy creation, evolution, and operation of the neighborhoods without the need for expensive central equipment and support, a Peer-to-Peer, multi agent, and negotiation based architecture has been designed and implemented to realize the control mechanism. Early evaluation has been based on user centered design and involvement, while final evaluation will be carried out using experiments and simulations based on one-year trials on two trial neighborhoods in Germany and Italy respectively.

Inferring Appliance Load Profiles from Measurements

Good demand side management in smart grids does not only depend on the amount of energy consumed by various appliances, but also on the temporal characteristics of the consumption, i.e. the load profile of the appliances. Representative load profiles can be used for predicting future energy consumption. However, a load profile is hard to characterise as it often depends on the operational conditions of the appliance when the measurements were taken. For instance the load profile of a washing machine will depend on the amount of cloths and the inlet water temperature. This paper presents a methodology for empirically obtaining the load profile from an ensemble of event driven traces of a stochastically varying mode of an appliance.

Intelligent Ant-Based Solution to the Application Service Partitioning Problem in a Grid Environment

This paper presents a decentralised multi-agent method based the metaphor of foraging intelligent ants, for solving the problem of application service partitioning onto the execution nodes of the grid environment such that all services of the application satisfy some minimum quality requirements. Fundamentally, this is an NP-hard problem. The proposed algorithms have been rigorously tested and evaluated through extensive simulations on randomly generated application services and grid environment. The results show that intelligent ants perform significantly better than what could be achieved with simple unintelligent random ants

From IoT big data to IoT big services

The large-scale deployments of Internet of Things (IoT) systems have introduced several new challenges in terms of processing their data. The massive amount of IoT-generated data requires design solutions to speed up data processing, scale up with the data volume and improve data adaptability and extensibility. Beyond existing techniques for IoT data collection, filtering, and analytics, innovative service computing technologies are required for provisioning data-centric and scalable IoT services. This paper presents a service-oriented design model and framework for realizing scalable and efficient acquisition, processing and integration of data-centric IoT services. In this approach, data-centric IoT services are organized in a service integrating tree structure, adhering to the architecture of many large-scale IoT systems, including recent fog-based IoT computing models. A service node in the tree is called a Big Service and acts as an integrator, collecting data from lower level Big Services, processing them, and delivering the result to higher level IoT Big Services. The service tree thereby encapsulates required data processing functions in a hierarchical manner in order to achieve scalable and real-time data collection and processing. We have implemented the IoT Big Services framework leveraging a popular cloud-based service and data platform called Firebase, and evaluated its performance in terms of real-time requirements.