Resource Allocation and Task Scheduling in the Cloud of Sensors

Abstract

The cloud of sensors (CoS) paradigm has emerged from the broader concept of Cloud of Things, and it denotes the integration of clouds and wireless sensor and actuator networks (WSANs). By integrating clouds with WSAN, some tasks initially assigned to smart sensors can be off-loaded to the cloud, thus benefiting from the huge computational capacity of these platforms. However, for time-critical applications, the high and unstable latency between the sensors and the cloud is not desirable. Besides low latency, WSAN applications usually require mobility and location-awareness properties, not often supported by current cloud platforms. Moreover, the indiscriminate off-loading of data/tasks from sensors to the cloud may lead to an overutilization of the network bandwidth while, in some cases, sensor-generated data could be locally processed and immediately discarded. To overcome these drawbacks of the integration between WSANs and the cloud, the edge paradigm emerges as a promising solution. Edge computing refers to enabling the computing directly at the edge of the network (for instance, through smart gateways and micro-data centers). Combining the paradigms of cloud/edge computing and WSANs in a three-tier architecture potentially leverages mutual advantages while posing novel research challenges. One of such challenges regards the development of solutions for performing resource allocation and task scheduling for CoS. Both edge and cloud paradigms strongly rely on the virtualization of physical resources. Therefore, resource allocation in CoS refers to the process of allocating instances of virtual nodes to perform the application requests (workload) submitted to the CoS, trying to meet as best as possible the requirements of applications, while respecting the constraints of the underlying physical infrastructure. Task scheduling denotes the process of selecting a group of physical nodes that are suitable for the execution, in a given order, of the various tasks necessary to meet an application request. The goal of this chapter is to overview the state of the art in the development of solutions for these two essential activities for the construction and efficient execution of CoS infrastructures.