Mohammad Hossain Heydari

On the Diameter of the Pancake Network

Then-dimensionalpancake network,Pn, has processors labeled with each of then! distinct permutations of lengthnand a connection between two processors when the label of one is obtained from the other by some prefix reversal. Each permutation is considered as a stack of different size pancakes. The well knownpancake problemconcerns the number,f(n), of prefix reversals required to sortnpancakes.We describe a (9/8)n+2 step sorting sequence for Gates and Papadimitrious stack ofnpancakes, ?n, used for establishing their lower bound, thus disproving the conjecture that (19/16)nsteps are required. Furthermore, we improve their lower bound by showingf(n)?(15/14)n. In fact, we define for eachn?0 (mod14) a stack of pancakes, ?n, and show that (15/14)n?f(?n)?(8/7)n?1.We show that ?In, the conjectured hardest stack of burnt pancakes, can be sorted in (3(n+1))/2 steps, for alln?3 (mod4) andn?23. If ?Inis indeed hardest, this implies that both the “burnt” and “unburnt” pancake networks of dimensionnhave diameter at most (3(n+1))/2.Values off(n), forn?11, were given previously. We note thatf(12)=14,f(13)=15, andf(19)?22.

Combinatorial optimization of multicast key management

There are numerous applications that require secure group communication. Much recent attention has been focused on secure multicasting over the Internet. When such systems are required to manage large groups which undergo frequent fluctuations in group membership, the need for efficient encryption key management becomes critical. This paper presents a combinatorial formulation of the multicast key management problem that is applicable not only to the specific problem of multicast key management, but also to the general problem of managing keys for any type of trusted group communication, regardless of the underlying transmission method between group participants. Specifically, we describe exclusion basis systems, show exactly when they exist, and demonstrate that such systems represent improvements over the current binary tree-based key management systems and other related systems.

Efficient Algorithms for Batch Re-Keying Operations in Secure Multicast

Secure multicast has a variety of applications in e-commerce, e-banking, command and control, video-on-demand, and other internet-based services. In this paper, we present algorithms to improve on the number of re-keying messages (overhead) needed to add and delete sets of users to a secure multicast group. We present upper and lower bounds on the number of re-keying bits to add or delete groups of users from secure multicast groups. We also show how to amortize the cost of group re-keying over time to avoid long periods of system overhead that can potentially block the transmission of desired data at times when large numbers of users simultaneously join or leave multicast sessions.

An intrusion-tolerant password authentication system

In a password-based authentication system, to authenticate a user, a server typically stores password verification data (PVD), which is a value derived from the users password using publicly known functions. For those users whose passwords fall within an attackers dictionary, their PVDs, if stolen (for example, through server compromise), allows the attacker to mount off-line dictionary attacks. We describe a password authentication system that can tolerate server compromises. The described system uses multiple (say n) servers to share password verification data and never reconstructs the shared PVD during user authentications. Only a threshold number (say t, t/spl les/n) of these servers are required for a user authentication and compromising up to (t-1) of these servers will not allow an attacker to mount off-line dictionary attacks, even if a users password falls within the attackers dictionary. The described system can still function if some of the servers are unavailable. We give the system architecture and implementation details. Our experimental results show that the described system works well. The given system can be used to build intrusion-tolerant applications.

Computing Cross Associations for Attack Graphs and Other Applications

Applications in information security, data mining, e-commerce, information retrieval and network management require the analysis of large graphs in order to discover homogeneous groupings of rows and columns, called cross associations. We show that finding an optimal cross association is NP-complete. Furthermore, we give a heuristic algorithm with an O(n4) running time for finding good cross associations

Algorithms and bounds for layer assignment of MCM routing

We present new algorithms for the layer assignment problem of multichip modules (MCMs). Our algorithms produce results that require between 70% and 25% of the number of layers required by the previous algorithms. We also present a new model for the problem that results in a better utilization of the routing area of the MCM, thus reducing the number of required layers even more. We provide lower and upper bounds on the performance of our algorithms which are tighter than the ones obtained before. Through our experimental results we show that the solutions obtained by our algorithms are close to the lower bounds. >

Towards trusted online dissemination of consumer information

Given the growing number of electronic services and their applications in many existing and evolving domains, such as e-tail and e-medicine, much recent attention has been focused on the collection, maintenance, and dissemination of personal information by electronic service providers (ESPs). Often, information is gathered without the active involvement of the consumers, who may be unaware of the nature of the data collected or the purposes for which these data will be used. Techniques such as cookies generally are implemented on the browsers without permission from the targeted consumer. In this paper, we propose a new framework for information dissemination to ESPs using secure communication of consumer-controlled information via trusted information distribution (TID) services. A TID service gathers information based on consumer consent and consumer-controlled profile then disseminates relevant information to targeted groups of ESPs. Using encryption and a group key management scheme, our communications mechanism securely and discriminately disseminates information to dynamic groups of ESPs that subscribe to the TID service.

Improved layer assignment for packaging multichip modules

The layer assignment problem plays an important role in packaging multichip modules, since the number of layers is directly related to the cost of the final product. In this paper, the authors propose a new model for the problem and a heuristic layer assignment algorithm based on the new model. The experimental results presented show that the solution provided by the algorithm is close to the lower bound.<<ETX>>

Stack and Queue Layouts for Toruses and Extended Hypercubes

Linear layouts play an important role in many applications including networks and VLSI design. Stack and queue layouts are two important types of linear layouts. We consider the stack number, s(G), and queue number, q(G), for multidimensional k-ary hypercubes and toruses. Heath, Leighton, and Rosenberg showed that d-dimensional ternary hypercubes have stack number &#8486;(N1/9), with N=3d nodes. Malitz showed that E edges implies stack number O(&#8730;E). For k-ary d-dimensional hypercubes, with N = kd vertices, Malitzs bound is O(kd/2). We improve this to 2d+1-3. The 2d+1-3 bound holds for arbitrary d-dimensional toruses. The queue number of d-dimensional k-ary hypercubes or toruses is bounded by O(d). Hence, Heath, Leighton, and Rosenberg exhibit an exponential tradeoff between s(G) and q(G) for multidimensional ternary hypercubes. Conversely, they conjectured that, for any G, q(G) is O(s(G)). We present a family {H} of modified multidimensional toruses and conjecture that q(H) is not O(s(H)).

Improved techniques for MCM layer assignment

Studies the layer assignment problem of multi-chip modules (MMCs) and presents algorithms for layer assignment of 2-terminal and multiterminal nets. Solutions obtained by our experimental results show a significant reduction in the number of plane-pairs required by our algorithms, in comparison with the previous algorithms. We improve the upper bound for multiterminal nets and show that the solutions obtained by our algorithms are close to the lower bounds.<<ETX>>