Yusuf Osmanlioglu

Delay minimizing user association in cellular networks via hierarchically well-separated trees

We study downlink delay minimization within the context of cellular user association policies that map mobile users to base stations. We note the delay minimum user association problem fits within a broader class of network utility maximization and can be posed as a non-convex quadratic program. This non-convexity motivates a split quadratic objective function that captures the original problems inherent tradeoff: association with a station that provides the highest signal-to-interference-plus-noise ratio (SINR) vs. a station that is least congested. We find the split-term formulation is amenable to linearization by embedding the base stations in a hierarchically well-separated tree (HST), which offers a linear approximation with constant distortion. We provide a numerical comparison of several problem formulations and find that with appropriate optimization parameter selection, the quadratic reformulation produces association policies with sum delays that are close to that of the original network utility maximization. We also comment on the more difficult problem when idle base stations (those without associated users) are deactivated.

Efficient approximation of labeling problems with applications to immune repertoire analysis

Labeling problems are finding increasing applications to optimization problems. They usually get realized into linear or quadratic optimization problems, which are inefficient for large graphs. In this paper we propose an efficient primal-dual solution, MLPD, for a family of labeling problems. We apply this algorithm to the analysis of immune repertoires, and compare it against our baseline approach based on refinement operators. We provide a comparative evaluation both in terms of accuracy and computational efficiency with respect to the baseline model, as well as to quadratic optimization.

Reducing parity generation latency through input value aware circuits

Soft errors caused by cosmic particles and radiation emitted by the packaging are an important problem in contemporary micropro-cessors. Parity bits are used to detect single bit errors that occur in the storage components. In order to implement parity logic, multiple levels of XOR gates are used and these XOR trees are known to have high delay. Many produced and consumed values inside a processor hold consecutive zeros and ones in their upper order bits. These values can be represented with less number of bits and hence are termed narrow. In this paper we propose a parity generator circuit design that is capable of generating the parity if the input value is narrow. We show that parity can be generated faster than a regular XOR tree implementation using our design for the values that can be represented using fewer bits.

Exploiting narrow values for faster parity generation

Purpose – The purpose of this paper is to reduce parity generation latency if the input value is narrow.Design/methodology/approach – Soft errors caused by cosmic particles and radiation emitted by the packaging are important problems in contemporary microprocessors. Parity bits are used to detect single bit errors that occur in the storage components. In order to implement parity logic, multiple levels of XOR gates are used and these XOR trees are known to have high delay. Many produced and consumed values inside a processor hold consecutive zeros and ones in their upper order bits. These values can be represented with less number of bits and hence are termed narrow. In this paper, a parity generator circuit design is proposed that is capable of generating parity if the input value is narrow. It is shown that the parity can be generated faster than a regular XOR tree implementation using this design for the values that can be represented using fewer bits.Findings – The proposed technique reduces the pari...

A primal-dual approach to delay minimizing user association in cellular networks

We study network utility maximization (NUM) within the context of cellular single user association (SUA) policies that map each mobile user (MU) to a single base station (BS) and make use of the generalized α-proportional fairness utility measure across downlink rates. Finding an exact solution to many such centralized user association problem is known to be NP-hard, so we are motivated to consider the integer relaxation of the SUA NUM problem. On this front, we provide separate characterizations of i) the fairness measures under which the SUA NUM problem integrality gap is exactly 1, and ii) the fairness measures yielding non-convex SUA NUM problem formulations. Next, we analyze the fairness measure corresponding to delay minimization and find a more natural linearization of the non-convex minimum delay SUA problem compared to our related previous work. We propose and construct a primal-dual algorithm to approximate the linearized minimum delay SUA problem. Our primal-dual algorithm is shown to achieve smaller performance gaps and runtimes over i) an intuitive baseline rounding algorithm applied to the linearized min delay SUA problem, as well as ii) two greedy heuristics that emphasize associations with minimal MU-BS distances and maximal downlink SINR ratios, respectively.

Multi-layer Tree Matching Using HSTs

Matching two images by mapping image features play a fundamental role in many computer vision task. Due to noisy nature of feature extraction, establishing a one-to-one matching of features may not always be possible. Although many-to-many matching techniques establishes the desired multi map between features, they ignore the spatial structure of the nodes. In this paper, we propose a novel technique that utilizes both the individual node features and the clustering information of nodes for image matching where image features are represented as hierarchically well-separated trees (HSTs). Our method uses the fact that non-leaf nodes of an HST represent a constellation of nodes in the original image and obtains a matching by finding a mapping between non-leaf nodes among the two HSTs. Empirical evaluation of the method on an extensive set of recognition tests shows the robustness and efficiency of the overall approach.

Unsupervised Motion Segmentation Using Metric Embedding of Features

Motion segmentation is a well studied problem in computer vision. Most approaches assume a priori knowledge of the number of moving objects in the scene. In the absence of such information, motion segmentation is generally achieved through brute force search, e.g., searching over all possible priors or iterating over a search for the most prominent motion. In this paper, we propose an efficient method that achieves motion segmentation over a sequence of frames while estimating the number of moving segments; no prior assumption is made about the structure of scene. We utilize metric embedding to map a complex graph of image features and their relations into hierarchically well-separated tree, yielding a simplified topology over which the motions are segmented. Moreover, the method provides a hierarchical decomposition of motion for objects with moving parts.

Biologically Inspired Attack Detection in Superpeer-Based P2P Overlay Networks

We present a bio-inspired mechanism that allows a peer-to-peer overlay network to adapt its topology in response to attacks that try to disrupt the overlay by targeting high-degree nodes. Our strategy is based on the diffusion of an “alert hormone” through the overlay network, in response to node failures. A high level of hormone concentration in a node induce that node to switch protocol. That leads to a self-organized modification of the entire overlay from a superpeer, scale-free layout, to a flatter network that is much less vulnerable to targeted attacks. As the hormone is metabolized with time, nodes switch back to the original protocol and reconstruct a superpeer overlay. We demonstrate and evaluate this mechanism on top of the peer-to-peer Myconet overlay, which is itself self-organized and bio-inspired.

Many-to-Many Matching under the l1 Norm

The problem of object recognition can be formulated as matching feature sets of different objects. Segmentation errors and scale difference result in many-to-many matching of feature sets, rather than one-to-one. This paper extends a previous algorithm on many-to-many graph matching. The proposed work represents graphs, which correspond to objects, isometrically in the geometric space under the l 1 norm. Empirical evaluation of the algorithm on a set of recognition trails, including a comparison with the previous approach, demonstrates the efficacy of the overall framework.