Research Interests

I am interested in all aspects of parallel and distributed systems from architecture to the application levels. However, I am mostly interested in the middleware layer between the application and the underlying operating system. The parallel and distributed community has been enjoying two recent very interesting application areas as the Mobile Networks and the Computational Grid.

• Mobile Ad hoc and Wireless Sensor Networks : Mobile ad hoc networks and wireless sensor networks are an excellent playground for distributed algorithm designers and graph theorists. They pose new problems for distributed systems as they have different requirements due to their dynamic topology. My recent work is centered around clustering algorithms, providing middleware functions such as clustering, distributed mutual exclusion, total order multicast as extensions to the operating system functions in mobile ad hoc networks, mostly based on graph theory and distributed algorithms. I am also interested in self-organization and self-stabilizing algorithms in wireless sensor networks. Some recent publications in this area are :

1. Dagdeviren, O., Erciyes, K., Cokuslu, D., Merging Clustering Algorithms for Mobile Networks, ICDCIT 2005, Springer Verlag, LNCS, (2005), to be published.

2. Marshall, G., Erciyes, K., Implementation of a Cluster Based Routing Protocol for Mobile Networks, ICCS2005, Springer Verlag, LNCS 3514, (2005), pp. 388-395.

3. Erciyes, K., Cluster-based Distributed Mutual Exclusion Algorithms for Mobile Networks , EUROPAR 2004, Springer-Verlag, LNCS 3149, (2004), pp. 933-940.

4. Erciyes, K., Marshall, G., A Cluster Based Hierarchical Routing Protocol for Mobile Networks, ICCS 2004, Springer-Verlag, LNCS 3045, (2004), pp. 528-537.

• The Computational Grid : Some of my recent work aims at providing load balancing and fault tolerance in a Computational Grid. A common problem with the Grid is that a computational node may fail which causes an application to crash. How can we provide fault tolerance mechanisms to prevent an application to crash ? One way of achieving this in a classical distributed system is to have active replication where copies of the application processes are run concurrently using finite state machines. How feasible and beneficial is it to apply the same principles to the Grid ? This is a current problem I am looking into. Also, I have been involved in the design and implementation of dynamic and distributed load balancing algorithms for Grids with some promising results. A recent publication is :

Erciyes, K., Payli, R., U., A Cluster-based Dynamic Load Balancing Middleware Protocol for Grids, EGC2005, Springer Verlag, LNCS 3470, (2005), pp. 805-812.

• Distributed, Real-time and Embedded Systems : I am mainly interested in distributed system functions realized as extensions of the operating system functions to the distributed environment such as distributed mutual exclusion, total order multicast and fault tolerance using group communications. My previous work involved scheduling and load balancing in distributed systems. If the word ‘real-time’ is added to any distributed system problem, one is confronted with a new class of problems which may be very different in nature than of the distributed system’s. Embedded systems are emerging as a discipline of its own with significantly different requirements than the distributed systems. I am looking into issues such as the functionality of the embedded operating systems. Some recent publications are :

1. Erciyes, K., Soysert, Z., Static Real-time Scheduling Algorithms using Graph Partitioning, ICCS2005, Springer Verlag, LNCS 3514, (2005), pp. 196-203.

2. Erciyes, K., Distributed Mutual Exclusion Algorithms on a Ring of Clusters, ICCSA 2003, Springer-Verlag, LNCS 3045, (2004), pp. 518-527.

3. Erciyes, K., Sahan, A., A Real-time Total Order Multicast Protocol, ICCS 2004, Springer-Verlag, LNCS 3036, (2004), pp. 357-364.

4. Sağlam, O, Dalkılıc, M., E., Erciyes, K., Design and Implementation of A Secure Group Communication Protocol on A Fault Tolerant Ring, Springer-Verlag, LNCS, (2003), pp. 802-810.

• Parallel Scientific Computing : In this case, I am interested in graph-theoretic approaches for parallel computing. One area I looked into is the graph partitioning problem and its application to the solution of sparse linear systems. We have also investigated fault tolerant routing algorithms in hypercube systems. Some related publications are :

1. Erciyes, K, Alp, A., Marshall, G, Serial and Parallel Multilevel Graph partitioning using Fixed Centers, Sofsem 2005, Springer-Verlag, LNCS 3381, (2005), pp.127-136.
2. Allahverdi, N, Kahramanli, S, Erciyes, K., A Fault Tolerant Routing Algorithm Based on Cube Algebra for Hypercube Systems , Journal of System Architecture (JSA), Vol.46-2, pp. 201-205. (2000).
3. Bakkaloglu, B., Erciyes, K., Koc, C.K., (1996), A Parallelization of Parlett's Algorithm for Functions of Triangular Matrices, Journal of Parallel Algorithms and Applications, Vol. 11/1-2, pp.61-69.
4. Allahverdi, N, Erciyes, K.(1998), Extension of Non-faulty Subcubes Set In a Faulty Hypercube Multiprocessor, Proc. of JCIS'98, Association for Intelligent Machinery, Oct. 23-28,1998,N.C.,USA.