Résumé: Fog computing has emerged as an extended version of the cloud infrastructure for providing latency-aware and scalable services for Internet of Things (IoT) devices. Adding the fog layer to the cloud computing paradigm improves the quality of service (QoS) of time-critical and delay-sensitive IoT applications. These applications consist of interconnected services that run on virtual machines (VM) or containers. Due to the resource-constrained nature of fog computing nodes and the interdependency of these VMs, the efficient placement of IoT applications has an impressive influence on the delay of IoT applications. Many IoT applications, such as industrial applications, health applications, smart cars, gaming applications, etc., are delay-sensitive. On the other hand, Distributed and heterogeneous features of fog nodes make it difficult to ensure the availability of these applications in fog environments. Unavailability in these applications (which is equivalent to infinite delay) can lead to irreparable loss. Therefore, various methods have been provided to ensure the availability of these applications. Using redundant VMs is an effective solution to ensure the availability of IoT applications. However, it is difficult to determine an efficient placement strategy for IoT applications that meet all of the required constraints of these applications in fog environments (such as resource constraints). As a result, this paper proposed a heuristic delay-efficient and availability-aware IoT application placement in fog environments called DAIP. DAIP tries to provide an application placement with low network delay to reduce applications delay. In addition, this method tries to provide an application placement with an acceptable level of availability. The DAIP is simulated using the iFogSim simulator. The simulations show that the DAIP method, in addition to ensuring the availability of applications, reduces network delay( and total delay) compared to the existing algorithms.

Langue: Anglais