Current TCP-friendly congestion control mechanisms such as those used in TFRC adjust the packet rate in order to adapt to network conditions and obtain a throughput not exceeding that of a TCP connection operating under the same conditions. In an environment where the bottleneck resource is packet processing, this is the correct behavior. However, if the bottleneck resource is bandwidth, and fl...

Network-based applications have evolved from simple web browsing to complex commercial transactions. This evolution has made a clear case for network QoS mechanisms, since the delays seen on the network often form the major component of an end-user's perceived delay. The key challenge in providing network QoS is the distributed and decentralized nature of the network resource. In this thesis, w...

Dynamic aggregate bandwidth over-reservation is a scalable approach for Quality of Service (QoS) control mechanisms, since surplus of reservation allows for admitting several flows without signaling the network. Our recent work, the Advanced Class-based resource Over-Reservation (ACOR), shows interesting results by significantly reducing QoS control signaling overhead with increased resource ut...

Current models and observations indicate that bacterial respiration should increase and growth efficiency (BGE) should decrease with increasing temperatures. However, these models and observations are mostly derived from data collected in temperate regions, and the tropics are under-represented. The aim of this work was to compare bacterial metabolism, namely bacterial production (BP) and respi...

A premier goal of resource allocators in virtualization environments is to control the relative resource consumption of the different virtual machines, and moreover, to be able to change the relative allocations at will. However, it is not clear what it means to provide a certain fraction of the machine when multiple resources are involved. We suggest that a promising interpretation is to ident...

We assume that only word pairs identified by human are available in a low-resource target language. The word pairs are parameterized by a bottleneck feature (BNF) extractor that is trained using transcribed data in a high-resource language. The cross-lingual BNFs of the word pairs are used for training another neural network to generate a new feature representation in the target language. Pairw...

As technology scales down, the interconnect for on-chip global communication becomes the delay bottleneck. In order to provide well-controlled global wire delay and efficient global communication, a packet switched Network-on-Chip (NoC) architecture was proposed by different authors [1][2]. In this paper, the NoC system parameters constrained by the interconnections are studied. Predictions on ...

Many visual tasks in modern personal devices such smartphones resort heavily to graphics processing units (GPUs) for their fluent user experiences. Because most GPUs for embedded systems are nonpreemptive by nature, it is important to schedule GPU resources efficiently across multiple GPU tasks. We present a novel spatial resource sharing (SRS) technique for GPU tasks, called a budget-reservati...

We study the polyhedral structure of a new model for various bottleneck resource allocation problems in which the resource consumption is determined by the activity with the maximum usage. The model, called the mixed vertex packing problem, is a generalization of the vertex packing problem having both binary and bounded continuous variables. We give valid inequalities and separation algorithms ...