Dear students,
Here is a list of topics for M.Tech projects. Please meet faculty and decide by March 25th. Unalloted students will be allotted topics after April 10. All faculty will take between 1 and 3 students, most taking 2. Some topics are with guide/co-guide. Prof MR has requested contact by email, initially, as he is traveling and also busy for some part of the time till March 25th. There will be small updates in this list in the coming few days.
The allocation scheme will be announced in a few days.
Narayan Rangaraj
------------------------------------------
Ashutosh Mahajan
1. Practical branching techniques for convex mixed-integer nonlinear optimisation.
The goal is to develop effective techniques for branching that can be used in the branch-and-bound algorithm for solving convex MINLPs. The project involves understanding existing techniques, development of new algorithms, software implementation and testing on benchmark instances.
2. Visualising branch-and-bound.
How do branch-and-bound trees look like in practice and how do they grow? What does their structure tell us about the nature of the problem (MILP/MINLP) and the method we are using to solve it? In this project, we will study these problems. Existing frameworks for visualising large trees will be examined. We will develop new ways to visualise progress of branch-and-bound algorithms.
3. (Jointly with Vishnu Narayanan) Optimal design of collimators for radiotherapy.
Existing MILP formulations for this problem are quite difficult to solve. We want to look at its polyhedral structure and develop tighter formulations. This project involves modeling the problem, studying the structure, discovering cutting planes and developing fast algorithms for separating fractional points.
4. (Jointly with Narayan Rangaraj) Cutting stock problem with setup times and other issues.
5. Any other well defined and interesting problem that a student wants to pursue.
V. Kavitha
1. Applications of Polling Systems
Polling systems are the queueing models in which a single server attends to more than one queue. When it completed service at one queue it switches over to another queue either in a given direction or randomly and it takes some non zero time to switch between two queues. Thus these are the queuing systems that do not satisfy work conservation law. Continuous polling systems are ones in which an arrival can occur anywhere on a cycle path in which the server is traveling periodically. Idea is to identify an interesting application and study the same by modeling it using an appropriate variety of the polling system. Applications can be from Railways or Manufacturing systems or Supply chain management.
2. Scheduling of Polling systems - When to switch optimally ?
Consider a Polling system with two independent queues deriving service from the same server. The question to be answered is when to switch between the two queues optimally, given the switching time, service time characteristics and arrival process distribution.
3. Mobility-based Power allocation in Cellular networks
Consider a cellular network catering to three varieties of users: a) stagnant or standing users; b) randomly wandering users; c) users traveling at high speeds on a street which passes through the cell. We analyzed each of these classes when they alone derive service the cellular network. The idea is to obtain the combined performance and then optimal power division among the classes.
4. Mathematical modeling of Railway traffic (joint project with Narayan Rangaraj)
The idea is to come up with an appropriate mathematical model which can fit the railway traffic and verifying the same using available data.
5. Simulation of a spatial queuing model/polling model.
Arrivals can occur anywhere in space. A server moves in a circular path placed inside the space. It attends any arrival when closest to it. The rate at which the service is offered depends upon the distance between the two. We would like to simulate such a model in Matlab and verify its estimated performance with corresponding theoretical one.
Vishnu Narayanan
1. Multileaf Collimator sequencing with integer programming (with Ashutosh Mahajan). This is a very hard combinatorial optimization problem arising from radiotherapy.
2. Stacking problems.
3. Some topic in transportation planning based on mutual interest.
K.S.Mallikarjuna Rao
1. Dynamic Kidney exchanges
2. Revenue sharing games in supply chains
3. Opinion dynamics
4. Games on networks and random graphs
5. Belief propagation and linear programming
6. Topic of mutual interest
7. Parking problems with incomplete information.
Jayendran Venkateswaran
1. Simulation based Optimization for complex manufacturing flow problems:
This involves implementation of simulation models and development of optimization module Plus simulation-optimization interface to help address a few problems such as (a) Production scheduling for Throughput Maximization (b) Buffer Capacity Minimization (c) Optimum Location for Buffers on the Mixed Model Assembly Line (d). Assembler Utilization Maximization etc. This project involves interaction with an industry for realistic data sets+ problems (involves travel). Good programming skills (say, C) needed. Early start on project will be emphasized.
2. Performance benchmarking and development of efficient metaheurisitcs for stochastic problems
Continuation of on going work, where idea is to benchmark the performance and understand the limitations of current meta heursitics in handling stochastic problems (unconstrained and constrained).. with eventual goal as to develop efficient meta heuristic that can give performance guarantees...
3. Impact of epidemics on healthcare system
Epidemic dynamics can be captured as a System Dynamics model, while hospital (queuing) activities modeling using discrete event simulation. This framework can be used to answer various question: What happens to hospital performance when epidemics occurs? How to time the hiring of additional staff to combat epidemics? What role does hospital play in the disease dynamics? Is it possible to consider multiple hospitals/clinics in a region? How to efficiently distribute vital medicines to affected regions?
4. Agent based free flow model of heterogenenous road traffic: The focus of this would be more on model development and its framework. A extrapolated general version could be modeling of human decision making.
5. GPU based discrete event simulation/ Parallelizing discrete event simulations
6. Simulation based optimization of scheduling/ routing/ pathing in Railways
7. Modeling and Analysis of sustainable supply chains: Considering carbon emissions or credits in facility locations & allocation models/ inventory management/ transportation/ bullwhip/ pricing/ production planning/ scheduling/ other SC decisions
8. Impact of input data analysis on output analysis
9. A neutral, and efficient interface for building manufacturing simulations using CMSD (Core manufactuing Simulation Data) framework with SimPy
10. Other topics related with, simulation modeling, simulation methodology, system dynamics, hybrid models, distributed systems, etc. Based on mutual interest.
N. Hemachandra
Topics in methodologies like Machine Learning, Markov decision models, Optimization, Game theory, Queueing models, etc., or applications of these and related methodologies to pricing and service level based problems arising from areas like Transportation systems, Power systems, Communications networks, Supply chains, etc.
Narayan Rangaraj
1. Fleet management in rail transport - petroleum rakes on Indian Railways
2. Facility planning and equipment reliability impact on railway traffic - with a sponsored project involving RDSO (Indian Railways), if it comes through.
3. Problems in time series analysis (with Prof A. Subramanyam - Mathematics) : one of two problems can be considered - (1) Co-integrated time series, where individual time series are not stationary (in the stochastic send), but where some linear combination is stationary and (2) forecasting in the near future - also called now-casting - and what techniques would be useful for this problem, including what data to consider.
A. Subash Babu
1. Decision problems in Retailing
2. Managing risks in Supply Chain
3. Resource Management in Projects
4. Decision problems in Service Sector
5. Decision problems in assembly based Manufacturing Systems
6. Any problem of mutual interest.
P.G.Awate
1. Scheduling of Freight Trains in railway networks with Single and Multiple tracks, using methods including Genetic Algorithms.
2. Scheduling in Multiple/Multi-mode Projects with Limited Resources: recent developments using Genetic algorithms.
3. Topic based on mutual interest.