- Optimization: Models, theory and algorithms
Mathematical programming models and discrete optimization continue to provide interesting techniques for analysis and decision making. Modelling, algorithms and computational methods in these areas provide for several areas of theoretical and applied research. The modelling and solutions of large network flow optimization problems have been some of the recent works in this area. These arise in logistics, transport and several other planning applications. Newer developments in integer programming and nonlinear programming provide powerful ways of solving large problems that arise in practice. These are also of interest to the group.
- Stochastic models
Stochastic models such as queueing models, etc., aim to capture the nature of systems that evolve dynamically over time in an random environment. In particular, queueing models provide a framework for sharing of resources for users that arrive over time. Some recent work explored in this area investigates the use of queueing models for design of service level based contracts between supply chain partners. Such contracts can aim for co-ordination in investment as well as usage of resources among various players in a supply chain. Contracts may as well aim optimal pricing and usage of resources while providing specified quality of service (QoS) levels. Another strand of work is the use of simulation based methods for parameter optimization as well as in differentiated pricing of network resources.
- Stochastic control
Broadly, this area concerns determination of optimal policies/strategies for control of systems that evolve in random environment.
- Markov decision process
Current interest in Markov decision process looks at sensitivity analysis of optimal policies w.r.t. various parameters like unit costs and transition probabilities of the model. A related topic is reinforcement learning algorithms of such models. The motivation of both these strands of work is to better our understanding regarding model uncertainty.
- Controlled diffusion
Current interest in control diffusion equations is to develop the theory of degenerate diffusion equations; a technically difficult topic. We also intend to look at applications in finance of such models.
- Markov decision process
- Simulation Modeling and Analysis
Simulation techniques are becoming essential in the modeling and analysis of complex systems. Research works are being carried out in the field of Discrete-Event Simulation, continuous System Dynamics simulation and Hybrid modeling. Primary areas of application are in manufacturing systems and supply chains, with potential applications in other areas. Framework and infrastructure for enabling distributed simulation is also developed using High Level Architecture (HLA) and Web Services.
Recent and ongoing works include production and distribution planning in supply chains using hybrid models; simulation-optimization; interface specifications and statistical data analysis in distributed simulation; control theoretic analysis of supply chains; dynamics of internal and external disturbances on supply chains behavior.
- Artificial Intelligence based methods
The interface of A.I. & O.R. has become increasingly interesting and active in areas such as: constraint satisfaction problems, stochastic search methods in context of many neural network learning algorithms including those for system prediction and control, neuro-dynamic programming and reinforcement learning. Ongoing work includes the application of reinforcement learning in robot navigation, neural net based optimization in scheduling problems.
- Game theory
There is renewed interest in Game theory these days as evinced by 2007 Nobel prize in economics for contributors to Mechanism design (Protocol engineering). Some of our work involves design of incentive compatible mechanisms for some problems that arise in supply chains, including co-ordination among partners. Another strand of work is to better understand broadly the topic of dynamic games. One aspect of our interest is average cost criterion in both deterministic and stochastic differential games. Another aspect is stochastic games wherein our interests are in approximate equilibria, games with stopping, etc.
- Logistics and Transportation
Transport and logistics forms a growing sector in Asia. These areas provide many interesting problems.
Studies of pricing and operations planning models for container, trucking and other fleet based transport operations are some areas of recent interest. This has led to work in the modeling of decision problems in the usage of shared resources, subject to service level constraints. A parallel area of work is the design of contracts in the logistics domain. Some work has been initiated in the area of air cargo operations.
Operational planning of transport operations, mode specific studies (Road, rail, air, sea), capacity planning, operations scheduling and routing, fleet and crew planning offer several rich problems for analysis. Ongoing work includes railway operations planning based on detailed simulations of train movements on the rail network. Many what-if analyses are possible, such as the impact of new schedules of trains, spacing of signal blocks, loop configuration at stations and level crossing control. Timetabling and rake allocation applications for suburban and long distance train services is another practical problem of considerable interest.
In the aviation sector, variants of crew planning and their scheduling (include rostering) and a variety of network flow based models for aircraft routing and deployment have been recent areas of research.
- Supply Chain Analysis and Inventory Planning
Some recent work involved exploring the role of contracts in facilitating co-ordination among different players such that the co-ordination results in supply chain optimality, apart from being beneficial to individual players. Some recent work includes investigating the stability and performance of various policies of incorporating returned goods in a supply chain by the paradigm of reverse logistics. Some previous work also included joint pricing and ordering decisions for supply chains involving perishable products.
- Financial Engineering
This new area includes modeling and pricing of derivative instruments, models from insurance and asset pricing, as well as the role of pricing and revenue management of services. Recent work included learning algorithms for portfolio management factoring in path dependent performance measures. A related work is in pricing of bonds.
- Optimization, Planning and Control in Manufacturing and Robotics
This area has seen very interesting new developments in the last two decades, such as: reactive scheduling systems, hierarchical and knowledge-based planning & control systems, search strategies for resource allocations, joint scheduling of machines and vehicles in flexible manufacturing systems, neuro fuzzy control in mobile robots, etc. Recently concluded work and ongoing work includes hierarchical planning for an asbestos sheet manufacturing plant, resource allocation strategies in multi level assembly system scheduling with setup times, neuro fuzzy control for navigation of mobile robot, and neural net based layout planning.
- Scheduling and ERP
With vast increases in computing power in last two decades, and also need for fast turnaround systems, there is increasing need for development and application of specialised algorithms in specific areas of application; e.g. vehicle scheduling problems, automated storage & retrieval systems, scheduling in machine tending robot systems. The growth of Enterprise Resource Planning systems and with distributed applications in today's e-business environment, there is a growing range of newer optimization problems that need to be addressed and studied. Ongoing work includes capacity-sensitive MRP and cyclic scheduling for a plant having two stage manufacturing & assembly system, dynamic scheduling priority strategies in flexible manufacturing systems, scheduling for machine tending robot.