The range and number of new applications and services are growing at an unprecedented rate. Computer networks need to be able to provide connectivity for these services and meet their constantly changing demands. This requires not only support of new network protocols and security requirements, but often architectural redesigns for long-term improvements to efficiency, speed, throughput, cost, and security. Networks are now facing a drastic increase in size and are required to carry a constantly growing amount of heterogeneous traffic. Unfortunately such dynamism greatly complicates security of not only the end nodes in the network, but also of the nodes of the network itself. To make matters worse, just as applications are being developed at faster and faster rates, attacks are becoming more pervasive and complex. Networks need to be able to understand the impact of these attacks and protect against them.
Network control devices, such as routers, firewalls, censorship devices, and base stations, are elements of the network that make decisions on how traffic is handled. Although network control devices are expected to act according to specifications, there can be various reasons why they do not in practice. Protocols could be flawed, ambiguous or incomplete, developers could introduce unintended bugs, or attackers may find vulnerabilities in the devices and exploit them. Malfunction could intentionally or unintentionally threaten the confidentiality, integrity, and availability of end nodes and the data that passes through the network. It can also impact the availability and performance of the control devices themselves and the security policies of the network. The fast-paced evolution and scalability of current and future networks create a dynamic environment for which it is difficult to develop automated tools for testing new protocols and components. At the same time, they make the function of such tools vital for discovering implementation flaws and protocol vulnerabilities as networks become larger and more complex, and as new and potentially unrefined architectures become adopted. This thesis will present the design, implementation, and evaluation of a set of tools designed for understanding implementation of network control nodes and how they react to changes in traffic characteristics as networks evolve. We will first introduce Firecycle, a test bed for analyzing the impact of large-scale attacks and Machine-to-Machine (M2M) traffic on the Long Term Evolution (LTE) network. We will then discuss Autosonda, a tool for automatically discovering rule implementation and finding triggering traffic features in censorship devices.
This thesis provides the following contributions:
1. The design, implementation, and evaluation of two tools to discover models of network control nodes in two scenarios of evolving networks, mobile network and censored internet
2. First existing test bed for analysis of large-scale attacks and impact of traffic scalability on LTE mobile networks
3. First existing test bed for LTE networks that can be scaled to arbitrary size and that deploys traffic models based on real traffic traces taken from a tier-1 operator
4. An analysis of traffic models of various categories of Internet of Things (IoT) devices
5. First study demonstrating the impact of M2M scalability and signaling overload on the packet core of LTE mobile networks
6. A specification for modeling of censorship device decision models
7. A means for automating the discovery of features utilized in censorship device decision models, comparison of these models, and their rule discovery
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/D8R5030T |
| Date | January 2017 |
| Creators | Jermyn, Jill Louise |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0024 seconds