This paper provides a survey and analysis of applications of Agent Based Simulation (ABS). A framework for describing and assessing the applications is presented and systematically applied. A general conclusion from the study is that even if ABS seems a promising approach to many problems involving simulation of complex systems of interacting entities, it seems as the full potential of the agent concept and previous research and development within ABS often is not utilized. We illustrate this by providing some concrete examples. Another conclusion is that important information of the applications, in particular concerning the implementation of the simulator, was missing in many papers. As an attempt to encourage improvements we provide some guidelines for writing ABS application papers.

This research investigates approaches to improve the performance of multi-agent based simulation (MABS) applications executed in distributed computing environments.  MABS is a type of micro-level simulation used to study dynamic systems consisting of interacting entities, and in some cases, the number of the simulated entities can be very large.  Most of the existing publicly available MABS tools are single-threaded desktop applications that are not suited for distributed execution.  For this reason, general-purpose multi-agent platforms with multi-threading support are sometimes used for deploying MABS on distributed resources.  However, these platforms do not scale well for large simulations due to huge communication overheads.  In this research, different strategies to deploy large scale MABS in distributed environments are explored, e.g., tuning existing multi-agent platforms, porting single-threaded MABS tools to distributed environment, and implementing a service oriented architecture (SOA) deployment model.

Although the factors affecting the performance of distributed applications are well known, the relative significance of the factors is dependent on the architecture of the application and the behaviour of the execution environment. We developed mathematical performance models to understand the influence of these factors and, to analyze the execution characteristics of MABS.  These performance models are then used to formulate algorithms for resource management and application tuning decisions.

The most important performance improvement solutions achieved in this thesis include: predictive estimation of optimal resource requirements, heuristics for generation of agent reallocation to reduce communication overhead and, an optimistic synchronization algorithm to minimize time management overhead.  Additional application tuning techniques such as agent directory caching and message aggregations for fine-grained simulations are also proposed.  These solutions were experimentally validated in different types of distributed computing environments.

Another contribution of this research is that all improvement measures proposed in this work are implemented on the application level.  It is often the case that the improvement measures should not affect the configuration of the computing and communication resources on which the application runs.  Such application level optimizations are useful for application developers and users who have limited access to remote resources and lack authorization to carry out resource level optimizations.

The computational Grid has become an important infrastructure as an execution environment for scientific applications that require large amount of computing resources.  Applications which would otherwise be unmanageable or take a prohibitively longer execution time under previous computing paradigms can now be executed efficiently on the Grid within a reasonable time.

Multi-agent based simulation (MABS) is a methodology used to study and understand the dynamics of real world phenomena in domains involving interaction and/or cooperative problem solving where the participants are characterized by entities having autonomous and social behaviour.  For certain domains the size of the simulation is extremely large, intractable without employing adequate computing resources such as the Grid. 

Although the Grid has come with immense opportunities to resource demanding applications such as MABS, it has also brought with it a number of challenges related to performance.  Performance problems may have their origins either on the side of the computing infrastructure or the application itself, or both. 

This thesis aims at improving the performance of MABS applications by overcoming problems inherent to the behaviour of MABS applications.  It also studies the extent to which the MABS technologies have been exploited in the field of simulation and find ways to adapt existing technologies for the Grid.  It investigates performance monitoring and prediction systems in the Grid environment and their implementation for MABS application with the purpose of identifying application related performance problems and their solutions.

Our research shows that large-scale MABS applications have not been implemented despite the fact that many problem domains that cannot be studied properly with only partial simulation.  We assume that this is due to the lack of appropriate tools such as MABS platforms for the Grid.  Another important finding of this work is the improvement of application performance through the use of MABS specific middleware.

Using simulation environments to support mobile distributed application development has many benefits. Since mobile devices can be modeled as autonomous intelligent agents, the multi agent based simulation (MABS) approach is useful for such applications. In this paper, we propose a MABS platform on the computational Grid for mobile distributed computing. A prototype of the proposed platform is implemented and its performance characteristics are studied The findings of the experiment show that a Grid based MABS platform can provide a scalable simulation environment for mobile distributed applications.

The computational Grid is an infrastructure which enables the execution of applications demanding huge computing resources. Hence, it can be the right environment for large-scale Multi-agent based simulation (MABS) applications. However, due to the nature of the Grid and the characteristics of MABS, achieving optimum performance poses a great challenge. Performance study of MABS applications is therefore a necessary undertaking which requires an understanding of these characteristics and the extent of their influence. Moreover, owing to the dynamicity and heterogeneity of the Grid, it is difficult to achieve performance gains without a middleware support for application deployment and dynamic reconfiguration. This paper presents a study of the key features of MABS applications that affect performance and proposes a supportive middleware to MABS platforms. Experiments show that the proposed middleware can bring performance improvement for MABS applications on the Grid.

The recent shift in machine learning towards the edge offers a new opportunity to realize intelligent applications on resource constrained Internet of Things (IoT) hardware. This paper presents a pre-trained Recurrent Neural Network (RNN) model optimized for an IoT device running on 8-bit microcontrollers. The device is used for data acquisition in a research on the impact of prolonged sedentary work on health. Our prediction model facilitates smart data transfer operations to reduce the energy consumption of the device. Application specific optimizations were applied to deploy and execute the pre-trained model on a device which has only 8 KB RAM size. Experiments show that the resulting edge intelligence can reduce the communication cost significantly, achieving subs-tantial saving in energy used by the IoT device.

A major requirement for Grid application developers is ensuring performance and scalability of their applications. Predicting the performance of an application demands understanding its specific features. This paper discusses performance modeling and prediction of multi-agent based simulation (MABS) applications on the Grid. An experiment conducted using a synthetic MABS workload explains the key features to be included in the performance model. The results obtained from the experiment show that the prediction model developed for the synthetic workload can be used as a guideline to understand to ,estimate the performance characteristics of real world simulation applications.

Distributed simulation has traditionally suffered from synchronization constraints that reduce the speedup significantly below the theoretical maximum. One solution approach are optimistic synchronization algorithms, allowing some nodes to advance in model time. In this paper, we present a novel algorithm that does not share the typical rollback costs of optimistic simulation, at the expense of restrictions on the design of simulation models. We have implemented that algorithm in an agent-based simulation framework, and evaluate its performance gains through measurements on actual simulation models.

Multi-agent based simulation (MABS) is a discrete event simulation technique used to study complex systems with entities having social and autonomous behavior. MABS applications are characterized by unpredictable execution behavior and high communication-to-computation ratio. In this paper, we propose an adaptation strategy to support efficient execution of large-scale MABS applications on typical Grid infrastructures. To achieve this objective, the behavior of MABS applications and the execution environment is investigated, in order to constantly obtain performance prediction models. These models will then be used to realize dynamic load balancing and resource allocation schemes. We discuss our basic approach, initial experimental results, the planned future research and an application of our research in the transportation and logistics simulation domain.

Agent platforms provide a framework for development and execution of parallel applications such as multi-agent based simulation (AMBS). However, these platforms have limitations to support large-scale MABS applications in practice. This paper aims at investigating and improving the performance of an agent platform with a AMBS workload in distributed environments. We carried out an experimental study with the JADE framework as our agent platform. The experiments show the performance characteristics of the workload and that JADE does not scale well due to message transport and agent directory service limitations. We propose solutions to overcome these performance bottlenecks and facilitate the efficient execution of AMBS in a distributed environment. Initial experimental results demonstrate the feasibility of the proposed approaches.