Network Framework for Multi-UAV Guided Ground Ad Hoc Network

Abstract

Unmanned aerial vehicles (UAVs) have gained lots of potential over the years, providing a vast range of applications in existing networks. UAVs have the ability to cooperate, and can fly autonomously or can be operated without human intervention, thus, provides a versatile, and a flexible implementation. Integration of these vehicles can solve various issues concerning both civilian activities as well as military operations. One of the possible integrations of these vehicles is with ad hoc networks. Ad hoc networks are low cost infrastructure based networks that operate on the ideology of forming intermittent networks on demand. Traditional ad hoc networks do not require any centralized device for data forwarding. Every node themselves act as transmitter, receiver, and router. Formation of collaborative network between the aerial nodes and the ground nodes provides a vast range of applications in areas of civilian and military activities. These collaborations can lead to formation of guidance systems that can be used to provide look ahead information to ground units. The work presented in this thesis considers the collaboration between these two different ad hoc units as a research aspect, and provides efficient strategies for enhanced transmission. In the initial phase, a cooperative framework is developed which forms a guidance system comprising of aerial and ground nodes operating in ad hoc mode. This framework forms a collaborative task oriented network, which carries search and tracking without any redundancy. The operating time of this framework is quite low. Also, it acts as a guidance system for ground nodes, thus, providing information of user stations on the ground. The proposed framework utilizes the Bayesian probabilistic model, neural networks, and Quaternion Kalman Filter for its successful operations. The proposed framework acts as a base for network formation between the aerial and the ground nodes.

In order to completely utilize the functions of the proposed framework, an inter-networking layer is incorporated into it. A new routing protocol based on the Fruit Fly Optimization (FFO) algorithm is proposed. This protocol utilizes the smell index property of fruit flies for selection of optimal route, maintenance, and rehabilitation. The proposed protocol provides an efficient approach for data/cognitive transmission in multi-UAV guided ad hoc networks. A detailed comparative analysis of the proposed protocol with some of the existing routing protocols is also presented. Further, this formation is improved by providing a service coordination facility over the proposed network framework. This service provisioning is achieved in three parts. In the first part, a middleware is proposed that provides an initial layout of layers functioning for efficient service coordination. In the second part, this layered middleware is subjected to cross layer formation for efficient service dissemination. In the third part, a complete model is presented that provides a 3-tier approach for efficient service coordination utilizing the initially proposed cooperative framework. The proposed 3-tier neural based service provisioning model provides a robust, reliable, and fault-tolerant connectivity between the ground ad hoc and the aerial ad hoc networks.

Utilizing the service enhancement, focus is driven towards the improvement of Quality of Service (QoS) to end users. Aiming at guaranteed QoS to end users, a queue scheduler is proposed for both ground as well as aerial assisted ad hoc networks. This approach is also developed in two parts. In the first part, a queue scheduler using the estimation and prediction strategies formulated over the Quaternion Kalman Filter is proposed. The proposed approach allows efficient estimation of network delays and bandwidth consumption. In the second part, this model is extended over UAV assisted network formations. This part also utilizes the features of existing Satisfied Importance Analysis (SIA) approach. The proposed approach is capable of providing guaranteed QoS to end users with improved connectivity.

Stabilization and broadcast storm are the other issues considered in this work. These issues hinder the network formation, and can cause serious deadlocks. Stabilization over UAVs provide continuous connectivity even in case of node failures. For stabilized connectivity, a self healing neural model is proposed that uses the concept of dummy neurons to provide stabilization against uncertain failures over networked UAVs. Broadcast storm causes redundancy and excessive utilization of resources for the same data. This is a serious issue in ad hoc formations between the aerial and the ground nodes. In order to eliminate this issue, a logical proximity graph bases parameter sensitivity routing (PSR) is proposed. The proposed approach not only solves the broadcast storm issue, but also improves the connectivity by lowering the chances of network partitioning.

In a cooperative network, security is one of the major issues. Here, security is defined in terms of localization, corridor privacy, and cognitive transfers. For this, a secure framework developed on the backbone of the initially proposed cooperative framework is proposed. This framework utilizes the alpha-beta-gamma filter to form a Teredo Tunnel for ad hoc network formation between the aerial and the ground nodes. The proposed framework withstands the network anomalies and faults, and proves robust and fault-tolerant by preventing any network hindrance during intrusions.