Distributed Run-time Management for Multi-agent System | Consortium for Embedded Systems | SIU

Abstract: Today’s prevalent solutions for modern multi-agent systems employ many processing inter-connected units leaving behind complex centralized approaches. Especially in modern automotive systems where high numbers of electronic components are employed. Navigation, car security, infotainment, travel information etc. are services highly depended on modern computing systems. In this proposed project, we couple the concept of multi-agent systems with run-time resource management techniques in order to develop a distributed framework for run-time management of multi-agent systems. The goal is to study already existing techniques, develop new solutions and implement them based on automotive constraints. The proposed framework will be based on the idea of local hypervisors and distributed agents in order to provide self-management functions while keeping system requirements.

Problem: The growing complexity of multi-agent systems, due to the high number of interconnected components and complex interactions among devices, will be unmanageable, and will obstruct the development of new services and applications. Distributed run-time management has been revealed as a key challenge to modern multi-agent systems and it has become prominent due to system’s dynamicity as resources can be added or removed from such environments at any time.

Rationale / Approach: A modern vehicle is considered as a complex multi-agent system due to the high numbers of electronic components employed. In modern automotive environments, one or many embedded electronic control units (ECU) are customized both on software and hardware having fixed functionality, each handling pre-specified inputs and services. Navigation, car security, infotainment, travel information, cruise control etc. are some examples of the services highly depended on ECUs. Thus, distributed run-time management and distributed ECUs prevail as the key technologies in order to address efficiently all problems generated by system's complexity. In this proposed project, we couple the concept of multi-agent systems with run-time resource management techniques in order to develop a distributed framework for runt-time management of multi-agent systems. The proposed framework will be based on the idea of local hypervisors and distributed agents in order to provide self-management functions while keeping system requirements.

Novelty: The novelty of the proposed framework will be threefold: (i) It will integrate run-time services for multi-agent systems in a distributed way; (ii) it will meet system requirements and (iii) it will provide a self-management functionality in the system.

Potential Member Company Benefits: The developed framework for distributed run-time management for multi-agent systems will be useful in many situations for the member companies. Firstly, new techniques regarding specific requirements such as power consumption and memory utilization will be evaluated. Also, the concept of multi-agent systems in modern automotive environment will be integrated and tested. Last, improvements and new solutions for self-management following the trends of a smart-car will be suggested.

Deliverables for the proposed year: I) Comprehensive report on the developed framework, II) Demo, with interconnected physical boards (e.g. beagleboard, raspberry pi over ethernet) of a simple scenario.

Milestones for the proposed year:
Quarter 1: Study of existing techniques and selection of the appropriate methodologies-algorithms
Quarter 2: Define architecture and hardware specifications for hypervisor and agents.
Quarter 3: Development of the aforementioned framework. First ideas about demo functionality
Quarter 4: Finalize development, demo and report.