| dc.description.abstract | The occurrence of a severe natural disaster causes loss of life and
destruction of properties. The overall criticality of the disaster depends
on the nature of the disaster and the physical characteristics of
the affected locations. In the aftermath of a natural disaster, multiple
emergencies often evolve at different geographical locations with casualties
and infrastructure damage. In post-disaster scenarios, responsible
authorities should initiate relevant disaster management activities
to mitigate the devastating effects of natural disaster. Resource allocation
is an integral part of the post-disaster activities.
In general, resource allocation deals with the issue of distributing necessary
resources to multiple users depending on their demand and the
availability of resources. It aims to achieve efficient and fair assignment
of limited resources. The devastation caused by a natural disaster
enforces the need for various critical resources in disaster-affected
locations to reduce the impact of the disaster. When adequate resources
are available, the problem of allocating resources becomes
trivial, and all the crisis locations can be fully satisfied in terms of
their resource requirements. However, if there is a scarcity of essential
resources after the simultaneous occurrence of multiple emergencies
at distinct geographical locations, providing resources to all those regions
and fulfilling their demands simultaneously becomes challenging.
In such situations, efficient decision-making is necessary to execute a
fair and socially agreeable allocation of resources to the affected locations.
One cannot rely on human-controlled decision-making since
it can have a bias for, or prejudice against, some of the disaster locations.
A fair and impartial approach to the allocation of resources can be implemented by designing an automated decision-making system.
This thesis proposes a game-theoretic framework which can form the
basis for such a system.
In this thesis, we develop a multi-event emergency management system
using a non-cooperative, single-stage, strategic form game model
to facilitate the allocation of resources to the respective disaster locations.
Each emergency event is assumed to occur at different locations
simultaneously, and some amount of resources are demanded by
each location to mitigate the impact of disasters. These locations are
represented as players in the game, which are assumed to play in a
self-interested manner with the other players to get an allocation of
scarce resources available at the resource station. However, it should
be noted that the disaster locations are not actively involved in playing
a game. It is a centralized decision-making executed by the responsible
disaster management authority, which implements the algorithm
designed using the game-theoretic framework to decide reasonable allocations
to the players. The authority assumes different allocations
to be the possible strategies of the players and arrive at a fair solution.
As a game utility, the authority imposes a non-monetary cost on each
player for obtaining a certain amount of resource units. The objective
of the proposed game is to derive socially acceptable strategies for an
effective and fair allocation of resources to the respective players. In
the thesis, it is established that the game model is unique in structure
and always possesses pure strategy Nash equilibria (PSNE). Each
PSNE consists of possible allocations to the players; hence, those can
be implemented by the disaster management authority as potential
allocation vectors.
As the resources needed during disaster management can be both
divisible and indivisible, we investigate the game for both types of resources.
Mathematical analysis shows that the existence of PSNEs is
independent of the nature of resources. The only difference it makes
is that in the case of indivisible resources, the players have a discrete set of strategies, and divisible resources make their strategy sets continuous.
It is also shown that the game-theoretic algorithm can be
used for any number of players or disaster locations at various stages
of resource allocations. The investigation is conducted using twoplayer,
three-player and n-player game models. Different case studies
are presented in the chapters of this thesis to validate the mathematical
results developed in this work and to indicate how this proposed
method can be helpful in practical disaster resource allocations. This
work also includes the statistical analysis of the game-theoretic algorithm
and the study of its computational complexity.
This thesis also includes a study on the preparedness and damage assessment
of a natural disaster using unmanned aerial vehicles (UAV).
Preparedness is a pre-disaster activity which is essential to build resilience
against natural disasters. Damage assessment is one of the
post-disaster activities which estimates the loss of human lives, properties,
and infrastructure. This phase is important to initiate the
response and recovery work after a natural disaster. These activities
become challenging and time-consuming when human effort is the
only option. In our study, we focus on the possible applications of
UAVs to make these activities speedy and effective. | en_US |
