Fog computing has emerged as a promising paradigm for distributed data processing, but managing numerous devices in fog domains is complex due to the scale of the infrastructure. To address this challenge, organizing fog devices into fog colonies allows independent management on a smaller scale. We present a genetic algorithm (GA) approach that utilizes hierarchical clustering to define the fog colony layout. The GA selects a subset of colony candidates from the dendrogram obtained with hierarchical clustering and optimizes the network communication time between users and applications and the execution time of algorithms that manage application placement in each colony. We deployed an NSGA-II, a multi-objective approach for GAs, to evaluate our proposal. Our experimental results demonstrate that combining a GA with hierarchical clustering improves both optimization objectives. We conducted nine experiment scenarios, varying the number of applications and fog devices. Our results show that even in the worst-case scenario, the GA’s results dominated the solutions obtained by two control algorithms after only 137 generations. Additionally, the number of genetic solutions and their homogeneous distribution in the Pareto front were satisfactory.

The increasing complexity of Compute Continuum environments calls for efficient resource optimization techniques. In this paper, we propose and evaluate three distributed designs of a genetic algorithm (GA) for resource optimization, within an increasing degree of distribution. The designs leverage the execution of the GA in the infrastructure devices themselves by dealing with the specific features of this domain: constrained resources and wide geographical distribution of the devices. For their evaluation, we implemented a benchmark case using the NSGA-II for the specific problem of optimizing the application placement, according to the guidelines of our three distributed designs. These three experimental scenarios were compared against a control case, representing a traditional centralized version of this GA algorithm, evaluating solution quality and network overhead. The results show that the design with the lowest distribution degree, which keeps centralized storage of the objective space, achieves comparable solution quality to the traditional approach but incurs a higher network load. The second design, which completely distributes the population between the workers, reduces network overhead but exhibits lower solution diversity while keeping enough good results in terms of optimization objective minimization. The second design demonstrates the highest overall efficiency in optimization performance and network cost. Finally, the proposal with a distributed population that only interchanges solutions between the workers’ neighbors achieves the lowest network load but with compromised solution quality.

Abstract We propose a framework designed to tackle a multi-objective optimization challenge related to the placement of applications in fog computing, employing a deep reinforcement learning (DRL) approach. Unlike other optimization techniques, such as integer linear programming or genetic algorithms, DRL models are applied in real time to solve similar problem situations after training. Our model comprises a learning process featuring a graph neural network and two actor-critics, providing a holistic perspective on the priorities concerning interconnected services that constitute an application. The learning model incorporates the relationships between services as a crucial factor in placement decisions: Services with higher dependencies take precedence in location selection. Our experimental investigation involves illustrative cases where we compare our results with baseline strategies and genetic algorithms. We observed a comparable Pareto set with negligible execution times, measured in the order of milliseconds, in contrast to the hours required by alternative approaches.

Abstract Osmotic computing encompasses emerging Cloud‐Internet of Things (IoT) computing paradigms, by featuring the possibility for application services to adapt into different functionally equivalent flavours, depending on the contextually available resources and on specific requirements of running applications. This article proposes a fully decentralised declarative framework that enables both application and infrastructure operators to declare management policies for the service instances and the nodes they manage, respectively. Policies are composed of a simple and well‐defined set of management operations, declared in Prolog, which trigger based on locally available contextual information on application requests and infrastructure resources. A prototype implementation of the framework is showcased and assessed via simulation over a lifelike Smart Campus use case with multiple applications, at increasing infrastructure sizes and number of mobile users. Experimental results show that the proposed management framework scales to large infrastructure sizes and suits the needs of multiflavoured Osmotic applications in dynamic deployment conditions, by improving the trade‐off between their response times and suitable service usage.

Abstract Orchestrating next-gen applications over heterogeneous resources along the Cloud-IoT continuum calls for new strategies and tools to enable scalable and application-specific managements. Inspired by the self-organisation capabilities of bacteria colonies, we propose a declarative, fully decentralised application management solution, targeting pervasive opportunistic Cloud-IoT infrastructures. We present a customisable declarative implementation of the approach and validate its scalability through simulation over motivating scenarios, also considering end-user’s mobility and the possibility to enforce application-specific management policies for different (classes of) applications.

Summary Fog computing architectures distribute computational and storage resources along the continuum from the cloud to things. Therefore, the execution of services or the storage of files can be closer to the users. The main objectives of fog computing domains are to reduce the user latency and the network usage. Availability is also an issue in fog architectures because the topology of the network does not guarantee redundant links between devices. Consequently, the definition of placement polices is a key challenge. We propose a placement policy for data replication to increase data availability that contrasts with other storage policies that only consider a single replica of the files. The system is modeled with complex weighted networks and topological features, such as centrality indices. Graph partition algorithms are evaluated to select the fog devices that store data replicas. Our approach is compared with two other placement policies: one that stores only one replica and FogStore, which also stores file replicas but uses a greedy approach (the shortest path). We analyze 22 experiments with simulations. The results show that our approach obtains the shortest latency times, mainly for writing operations, a smaller network usage increase, and a similar file availability to FogStore.

Summary Fog computing aims at extending the cloud towards the Internet of things so to achieve improved quality of service and to empower latency‐sensitive and bandwidth‐hungry applications. The fog calls for novel models and algorithms to distribute multiservice applications in such a way that data processing occurs wherever it is best placed, based on both functional and nonfunctional requirements. This survey reviews the existing methodologies to solve the application placement problem in the fog, while pursuing three main objectives. First, it offers a comprehensive overview on the currently employed algorithms, on the availability of open‐source prototypes and on the size of test use cases. Second, it classifies the literature based on the application and fog infrastructure characteristics that are captured by available models, with a focus on the considered constraints and the optimized metrics. Finally, it identifies some open challenges in application placement in the fog.

Fog computing extends the cloud to where things are by placing applications closer to the users and Internet of Things devices. The placement of those applications, or their services, has an important influence on the performance of the fog architecture. Improving the availability and the latency of the applications is a challenging task due to the complexity of this type of distributed system. In this paper, we propose a service placement policy inspired by complex networks. We are able to increase the service availability and the quality of service (QoS) satisfaction rate by first mapping applications to communities of fog devices and then transitively placing the services of the applications on the fog devices of the community. The underlying idea is to place as many interrelated services as possible in the devices closest to the users. We compare our solution with an integer linear programming approach, and the simulation results show that our proposal obtains improved QoS satisfaction and service availability.

Resource management is a key factor in the performance and efficient utilization of cloud systems, and many research works have proposed efficient policies to optimize such systems. However, these policies have traditionally managed the resources individually, neglecting the complexity of cloud systems and the interrelation between their elements. To illustrate this situation, we present an approach focused on virtualized Hadoop for a simultaneous and coordinated management of virtual machines and file replicas. Specifically, we propose determining the virtual machine allocation, virtual machine template selection, and file replica placement with the objective of minimizing the power consumption, physical resource waste, and file unavailability. We implemented our solution using the non-dominated sorting genetic algorithm-II, which is a multi-objective optimization algorithm. Our approach obtained important benefits in terms of file unavailability and resource waste, with overall improvements of approximately 400 and 170 percent compared to three other optimization strategies. The benefits for the power consumption were smaller, with an improvement of approximately 1.9 percent.

The performance of web caches, in Web Content Management Systems, can be improved by assembling only some of the content elements of a web page in the application server, and finishing the assembling process in the cache proxy. Due to this, the cache is able to manage parts of the web page instead of whole pages, which improves its performance. We propose an algorithm based on decision trees and obtained in a training process to create content fragmentation designs. Data mining is used in the training phase. Inputs of the classification algorithm must be monitored from the system producing small overheads. The paper contribution are the validation of: the use of classification system to self-adapt content fragmentation designs to improve the web performance, the parameters set to be used as inputs of the decision tree and finally, the suitability of using decision trees to represent and implement, in the classification system, the previous extracted knowledge. All these aspects are validated by experimental results extracted from a test-bed.

Companies and web sites store information about the user behaviour which is not shared with other people. A tool for public and free sharing of this information has been developed. This tool is installed as a Firefox Add-on and it stores, in a central system (Web Mining Service), the gathered information in the browser of the client.

We present a Web system architecture using ontologies to improve the behavior of the system from the performance viewpoint. Since Web system performance indexes depend on state and parameter values on runtime period, the proposed system configuration will change during this period. In order to perform this change, the Web system is monitorized and gathered information stored into a knowledge base. We also model the performance of the different Web system elements intervening in the configuration using the knowledge base expressed by means of ontologies. An example of the use of this ontology in cache tier is also presented. We propose the use of performance reasoners to change the configuration during runtime period based on the information supplied from the knowledge base.