M. Todd Gardner

Evaluating geographic vulnerabilities in networks

In wireless ad-hoc and wireline networks used for search and rescue, military operations, and emergency communications; many failure modes are geographic in nature. They include jammers, explosions, enemy attacks, terrain issues, and natural causes like floods, storms, and fires. This paper proposes two methods to gain valuable insights into the physical topography and geographic vulnerabilities of networks. The 2-Terminal method and All-Terminal method find areas that given a threat of a certain radius can disconnect either the source and destination pair or any component of the network respectively. We believe that these methods could be used to optimize network node selection, placement and design. To be tractable, both methods incorporate innovative search techniques to use the size of the threat to reduce the complexity of the search.

Optimised heuristics for a geodiverse routing protocol

We propose two heuristics for solving the path geodiverse problem (PGD), in which the calculation of a number of geographically separated paths is required. The geodiverse paths can be used to circumvent physical challenges such as large-scale disasters in telecommunication networks. The heuristics we propose for solving PGD have significantly less complexity compared to the optimal algorithm we previously used while still performing well by returning multiple geodiverse paths for each node pair. The geodiverse paths contribute to providing resilience against regional challenges. We present the GeoDivRP routing protocol with two new routing heuristics implemented, which provide the end nodes with multiple geographically diverse paths and demonstrates better performance compared to OSPF when the network is subject to area-based challenges.

Analysing GeoPath diversity and improving routing performance in optical networks

With the increasing frequency of natural disasters and intentional attacks that challenge telecommunication networks, vulnerability to cascading and regional-correlated challenges is escalating. Given the high complexity and large traffic load of optical networks, these correlated challenges cause substantial damage to reliable network communication. In this paper, we propose a network vulnerability identification mechanism and study different vulnerability scales using real-world optical network data. We further propose geographical diversity and incorporate it into a new graph resilience metric cTGGD (compensated Total Geographical Graph Diversity), which is capable of characterising and differentiating resiliency levels among different optical fibre networks. It is shown to be an effective resilience level indicator under regional network challenges or attacks. We further propose two heuristics for solving the path geodiverse problem (PGD) in which the calculation of a number of geographically separated paths is required. Geodiverse paths can be used to circumvent physical challenges such as large-scale disasters in telecommunication networks. We present the GeoDivRP routing protocol with two new routing heuristics implemented, which provides the end nodes with multiple geographically diverse paths. Our protocol demonstrates better performance compared to OSPF when the network is subject to area-based challenges. We have analysed the mechanism by which the attackers could use to maximise the attack impact with a limited budget and demonstrate the effectiveness of restoration plans.

Using Multi-Topology Routing to improve routing during geographically correlated failures

During large geographic events in networks, the routing churn that occurs has been shown to cause significant impacts in routing stabilization following the event. This work proposes a set of algorithms that is based on Multi-Topology Routing (MTR) for pre-planning for geographically correlated failures. Thus, in the event of a failure, our approach, Geographic MTR, switches to virtual topologies that reduce the impact of routing changes that can result in dropped connections until a new link state and shortest path trees can be established. We propose two algorithms to generate virtual topologies, Geographic Coverage MTR (gcMTR) and Geographic Targeted MTR (gtMTR). The first method, gcMTR, is to create virtual topologies taking a network wide coverage approach, while gtMTR is a targeted approach that can be used in anticipation of a specific event where knowledge of that event exists. A third algorithm proposed in this work specifies a way to detect a geographic event and select a topology to use. We evaluated our approach on two network topologies and observed that the number of connections that are dropped during a geographic event can be reduced significantly, thereby reducing the impact to the non-affected part of the network.

A Geographic Multi-Topology Routing approach and its benefits during large-scale geographically correlated failures

Large-scale geographical events can significantly disrupt network services. In particular, the routing churn that occurs during such large-scale events has been shown to cause significant impact in route stability and transient behavior. We take a Geographic Multi-Topology Routing (gMTR) approach for pre-planning of geographically correlated failures. Thus, in the event of a failure, the gMTR approach switches to a virtual topology that reduces the impact of routing changes that can result in dropped connections until new paths can be established. Two algorithms are proposed to generate virtual topologies, Geographic Coverage MTR (gcMTR) and Geographic Targeted MTR (gtMTR). The first method, gcMTR, is to create virtual topologies taking a network wide coverage approach for which we consider taking both a circular coverage approach and a hexagonal coverage approach. gtMTR, on the other hand, is a targeted approach that can be used in anticipation of a specific event where the knowledge of the impending event is available. We propose another algorithm that specifies a way to detect a geographic event and select a topology to use. We evaluated our approach on two network topologies and observed that the number of connections that are dropped during a geographic event can be reduced significantly using our gMTR approach, thereby reducing the impact to the non-affected part of the network. We performed an analysis of the topology size versus the disaster size, topology location versus disaster location, and general density of the topology. Finally, a simulation model of the larger topology is used to study the effects of geographically correlated failures both with gcMTR and using default topologies. This provides a way to assess the gains from using gMTR to mitigate the impact of large scale geographic impacts.

Finding geographic vulnerabilities in multilayer networks using reduced network state enumeration

Despite advancements in the analysis of networks with respect to geographic vulnerabilities, very few approaches exist that can be applied to large networks with varied applications and network measures. Natural and man-made disasters as well as major political events (like riots) have kept the challenges of geographic failures in networks in the forefront. With the increasing interest in multilayer and virtual networks, methods to analyze these networks for geographic vulnerabilities are important. In this paper, we present a state space analysis method that analyzes multilayer networks for geographic vulnerabilities. It uses either the inability to provision an upper layer service and/or increased costs to provision as the criteria for network failure. Mapping techniques for multilayer network states are presented. Simplifying geographic state mapping techniques to reduce enumeration costs are also presented and tested. Finally, these techniques are tested on small and extremely large networks.

Creating network resilience against disasters using Service Level Agreements

Building networks that are resilient to natural disasters and other geographically correlated events involves both proactive network design and remediation after the event. Network provisioning is one of the steps used to provide redundancy and diversity in networks that are key elements in resilient network design. Once the disaster has occurred, remediation may include rerouting of the most important services affected by the event. Both components involve understanding the quality of service that may be affected, knowing which services are critical, and having an understanding of the geographic events that may impact the network. Furthermore, if resources are limited following the disaster, how do we select which services to reroute? In this work, we propose to use Service Level Agreements (SLA) to provide much of the information necessary to solve the provisioning and rerouting problems. On the provisioning side, the SLAs will provide the required system response time, availability and survivability of a service. On the remediation side, the SLA can provide the priority of a service to restore and reroute. We present mixed-integer linear programming (MILP) models and heuristics to show how the most important services are provisioned and rerouted to be available in the event of a geographic disaster. Our approaches are found to be more efficient than other priority-based approaches.

Determining Geographic Vulnerabilities Using a Novel Impact Based Resilience Metric

Various natural and man-made disasters as well as major political events (like riots) have increased the importance of understanding geographic failures and how correlated failures impact networks. Since mission critical networks are overlaid as virtual networks over a physical network infrastructure forming multilayer networks, there is an increasing need for methods to analyze multilayer networks for geographic vulnerabilities. In this paper, we present a novel impact-based resilience metric. Our new metric uses ideas borrowed from performability to combine network impact with state probability to calculate a new metric called Network Impact Resilience. The idea is that the highest impact to the mission of a network should drive its resilience metric. Furthermore, we present a state space analysis method that analyzes multilayer networks for geographic vulnerabilities. To demonstrate the methods, the inability to provision a given number of upper layer services is used as the criteria for network failure. Mapping techniques for multilayer network states are presented. Simplifying geographic state mapping techniques to reduce enumeration costs are also presented and tested. Finally, these techniques are tested on networks of varying sizes.

Using QNA to Evaluate Parameter Tuning in Mission Critical SOA Networks

One of the limitations of using Service Oriented Architecture (SOA) in mission critical applications (like Air Traffic Control) is the ability to specify stringent Quality of Service (QoS) parameters for individual services. Even determining which Message Broker optimizations to utilize to achieve high QoS is difficult. This research is utilizing the concepts introduced by the Queuing Network Analyzer (QNA) [1] to model the performance of SOA networks with features like message chunking, message exchange patterns (MEP), multi-layer structure and multiple classes of products. Non-exponential service and arrival distributions are investigated. As this is work in progress, we have not completed the simulation models to support the analytic models presented in this paper.

Provisioning dynamic and critical demand structures for geographically correlated failures: Using multi-disaster approach

During disasters and other geographically correlated failures, the local telecommunications infrastructure undergoes a unique set of challenges. First, the telecommunications infrastructure is typically damaged reducing the capability to sustain the normal level of demands on the networks. Second, the demands on the telecommunications infrastructure increase typically by an order of magnitude. This is both for critical demands and non-critical demands. Third, critical demands, which may now be used to support disaster first responder and recovery efforts take on a crucial role. In this work, we propose multi-situational linear programming techniques to create disaster modes that take advantage of the non-coincidence of disasters in different geographic areas simultaneously reducing cost and improving restoration of demands during disasters. We also propose efficient heuristics to provision for disasters at a lower cost than standard provisioning with 1 + 1 redundancy. Additionally, we use diverse routing techniques to reduce the likelihood of damage to critical services. The restoration of demands during disasters also presents challenges for network provisioners. We present both heuristic and linear programming methods to assist with restoration of services in stressed network environments giving critical demands priority. Finally, service level agreements (SLAs) are used to provide a framework for these concepts.