Justin Yates

A constrained binary knapsack approximation for shortest path network interdiction

A modified shortest path network interdiction model is approximated in this work by a constrained binary knapsack which uses aggregated arc maximum flow as the objective function coefficient. In the modified shortest path network interdiction problem, an attacker selects a path of highest non-detection probability on a network with multiple origins and multiple available targets. A defender allocates a limited number of resources within the geographic region of the network to reduce the maximum network non-detection probability between all origin-target pairs by reducing arc non-detection probabilities and where path non-detection probability is modeled as a product of all arc non-detection probabilities on that path. Traditional decomposition methods to solve the shortest path network interdiction problem are sensitive to problem size and network/regional complexity. The goal of this paper is to develop a method for approximating the regional allocation of defense resources that maintains accuracy while reducing both computational effort and the sensitivity of computation time to network/regional properties. Statistical and spatial analysis methods are utilized to verify approximation performance of the knapsack method in two real-world networks.

Optimizing social media message dissemination problem for emergency communication

Derive mathematical formulation to model social media message dissemination.Large experimental design using random and real-world (crawled) social media networks.Observe strong correlations between message reception time and node degree.Node delay consistently affects message reception time.Random networks over-estimate message dissemination and model performance. Social media is increasingly being used as a communication bridge between government, emergency responders and managers, and the general public in extreme events. Passing information through social media channels enables individuals to send and receive content in real-time and without limitation of location and geography. While the use of social media in extreme event situations has become prevalent, there is often little strategy involved in message dissemination and too little understanding of the effects that underlying online social networks have on message distribution. In this study, we introduce a formal model for social media message dissemination in social networks through time. Our proposed model includes emphasis on single and multiple message scenarios and examines key communication characteristics in the development of more intentional and targeted social messaging strategies. We present a detailed experimental design on randomly generated networks and real-world sub-networks of the Twitter social graph and discuss our findings. We also include a Tabu Search procedure for solving single-message problem and discuss its potential value for large-scale problems in real-world applications.

A length-based, multiple-resource formulation for shortest path network interdiction problems in the transportation sector

Abstract This paper analyzes a variation of the shortest path network interdiction problem for homeland security scenarios pertaining to attacks on critical infrastructure and key resources that use highways in the transportation sector as conduits for gaining proximity to targets. The model represents a static Stackelberg game and may be formulated as a bi-level mixed integer program with two players: an attacker and a defender. Using highway segments as arcs, a set of predetermined highway entry points and a target set, the attacker seeks the path of maximum non-detection between any entry and target node. The defender impacts the minimum value of this maximum non-detection path through the allocation of a limited number of defense sensors that reduce the non-detection probabilities for arcs that fall within the range of influence of a sensor. Two types of sensors, static and dynamic sensors, are available to the defender and separate influence functions model their respective effects on arc non-detection. A geographic information system is used to collect, store and process network information and sensor influence information stored in a relational database. The results of the problem formulation are analyzed in a case study involving a California highway sub-network. The case study also examines the effects of sensor parameters, budget levels and target sets on the solutions that are obtained.

Assessing the impact of vulnerability modeling in the protection of critical infrastructure

This paper examines the impact of arc metrics on the computational performance and spatial similarity in network interdiction modeling. Computational impact is measured in the number of iterations and total time required to reach an optimal solution. A combination of spatial analytical tools is offered as a methodology to assess the similarity in defense resource allocation when applying different arc interdiction metrics. An experimental design was devised and implemented using two real-world sub-networks of the Los Angeles County roadway system. This paper shows that arc metric selection has a limited effect on the spatial allocation of defense resources though metric choice does directly impact computation time. These results have direct implications to public policy and decision-making by enabling a modeler to increase his/her situational awareness and also their confidence in resource allocation decisions by selecting metrics that will improve their solution capabilities.

Identifying key parameters and trends in civil violence: a sub-regional, agent-based simulation approach using GIS

This work extends a popular agent-based simulation model of civil violence and analyses the effects of these extensions within the test-case region of Iran. First, a macro-level model of civil violence using geographic information science methods and real-world transportation network data is modelled and introduced. A detailed experimental design analyses the sensitivity of the modelled outbreak of civil violence on road network structures. Next, we use individual agent location to develop a personal legitimacy value for each agent of the system and model this legitimacy as a function of agent movement through the region. The resulting models indicate that the occurrence of civil violence in the derived simulations is very sensitive to network composition and connectivity of a given sub-region and identify a small number of behavioural outbreak trends to which sub-regions can be classified. We also show that regional legitimacy can have a marked effect on central authority agent distribution and movement.

Network Interdiction Methods and Approximations in a Hazmat Transportation Setting

The United States transportation system is an extensive and integrated component in the eight key infrastructures upon which the livelihood of the U.S. is dependent (Department of Homeland Security 2009). The accessibility and mobility enabled through open use of the transportation system is a vital and necessary freedom which contributes to the fluidity of the American environment. The transportation system is expansive and heavily utilized with an average of over 2 billion daily vehicle-miles of travel (nearly twice as much travel since the early 1980s) on the roughly 4 million miles of paved roadway, nearly 47,000 miles of Interstate highway, 600,000 bridges and 366 U.S. highway tunnels over 100 m (Texas Transportation Institute 2011; Transportation Security 2012). Travelers and shippers may also choose to utilize more than 300,000 miles of freight rail, nearly 10,000 miles of urban and commuter rail systems, or connect between 500 commercial-service and 14,000 general aviation airports (Transportation Research 2002).

Soccer Analytics Using Touch-by-Touch Match Data

This paper discusses several soccer analytics directions exploiting detailed ball touch data from a soccer game. The topics discussed include visualizing team formations and quantifying territorial advantage; determining the network-based structural properties of team play, and computing the importance of individual players for the team interactions. The proposed ideas are illustrated using the data from a real-life Barclays Premier League game, which was made available by StatDNA.