Random Access in Capillary Machine Type Networking: Modeling, Analysis and Optimization
Abstract
Networked infrastructures of capillary nature have lately been in emergence for a particular variety of machine type communication involving miniaturized sensors and actuators as densely located end-nodes. As an enabler of pervasive systems, these networks facilitate interplay among a wide range of applications such as ambient intelligence, safety and security, smart grid/energy services and utility control. With largely event-driven packet traffic, random access schemes are a logical choice at the medium access control layer in these networks. Yet, the large scale of the networks within the link layer neighborhood, the low- power nature of most of the end-nodes, and the potential imperative to prioritize critical traffic pose fundamental challenge to random access regarding efficiency and flexibility.
This thesis addresses the challenge considering carrier sense multiple access (CSMA) based schemes from an analytic modeling approach. Leveraging the infrastructure-based setting expected in the access network, investigations have been undertaken on assessing ways for priority-provisioning and optimizing performance. First, a conventional variant of the CSMA protocol embodying a fixed provision for priority is modeled with stochastic formulation and the throughput and packet service time performance are evaluated. In a similar track, a CSMA protocol is modeled in tractable form to determine throughput optimizing condition in conjunction with granting arbitrary degrees of priority. Flexible prioritizing with optimality for a related CSMA scheme is also studied with an immediately enforcing mechanism for differentiation. Second, accounting for intermittent packet traffic within the access network, an algorithmic solution is produced for the coordinator to estimate time-varying random access contention, leading to high CSMA utilization on-the-fly with fairness among the nodes and low temporal dispersion. Lastly, CSMA has been modeled and analyzed considering multi-packet reception capacity at the underlying physical layer, leading to the determination of a suitable backoff scheme for the task.
The solutions are a step towards instilling self-organizing capacity into the multifaceted, software-defined systems in emergence for managing operations of the related networks.
Description
Keywords
Computer Science, Engineering--Electronics and Electrical
Citation
Kazi, A. (2017). Random Access in Capillary Machine Type Networking: Modeling, Analysis and Optimization (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/27522