Sherif F. Fahmy

Response time analysis of software transactional memory-based distributed real-time systems

We consider distributed real-time systems where concurrency control is managed using software transactional memory (or STM). For such a method we propose an algorithm to compute an upper bound on the response time. We compare the result of the proposed algorithm to a simulation of the system being studied in order to determine its efficacy. The results of our study indicate that it is possible to provide timeliness assurances for systems programmed using STM.

On bounding response times under software transactional memory in distributed multiprocessor real-time systems

We consider multiprocessor distributed real-time systems where concurrency control is managed using software transactional memory (or STM). For such a system, we propose an algorithm to compute an upper bound on the response time.The proposed algorithm can be used to study the behavior of systems where node crash failures are possible. We compare the result of the proposed algorithm to a simulation of the system being studied in order to determine its efficacy. The results of our study indicate that it is possible to provide timeliness guarantees for multiprocessor distributed systems programmed using STM.

Chitosan-tripolyphosphate nanoparticles: Optimization of formulation parameters for improving process yield at a novel pH using artificial neural networks

At a novel pH value of the polymeric solution (6.2), variable chitosan (Cs) and sodium tripolyphosphate (TPP) concentrations and mass ratios were optimized to improve the process yield without undesirable particle flocculation. Prepared formulations were characterized in terms of particle size (PS), zeta potential (ZP) and percentage yield (% yield). Artificial neural networks (ANN) were built up and used to identify the parameters that control nanoparticle (NP) size and yield, in addition to being tested for their ability to predict these two experimental outputs. Using these networks, it was found that TPP concentration has the greatest effect on PS and% yield. The most optimum formulation was characterized by a notable process yield reaching 91.5%, a mean hydrodynamic PS 227 nm, ZP+24.13 mv and spherical compact morphology. Successful Cs-TPP interaction in NP formation was confirmed by both Fourier transform-infrared spectroscopy (FT-IR) and differential scanning calorimetry (DSC). This study demonstrated the ability of ANN to predict not only PS of the formed particles but also NP% yield. This may have a great impact on Cs-TPP NPs preparation and can be used to customize the required target formulations.

On Collaborative Scheduling of Distributable Real-Time Threads in Dynamic, Networked Embedded Systems

Some emerging networked embedded real-time applications have relatively long reaction time magnitudes-e.g., milliseconds to minutes. These longer execution time magnitudes allow opportunities for more computationally expensive scheduling algorithms than what is traditionally considered for device-level real-time control sub-systems. In this paper, we review recent research conducted on collaborative scheduling algorithms in such systems that are subject to dynamic behavior such as transient and sustained resource overloads, arbitrary activity arrivals, and arbitrary node failures and message loss. We show that collaborative scheduling algorithms have an advantage over non-collaborative scheduling algorithms.

Fast Scheduling of Distributable Real-Time Threads with Assured End-to-End Timeliness

We consider networked, embedded real-time systems that operate under run-time uncertainties on activity execution times and arrivals, node failures, and message losses. We consider the distributable threads abstraction for programming and scheduling such systems, and present a thread scheduling algorithm called QBUA. We show that QBUA satisfies (end-to-end) thread time constraints in the presence of crash failures and message losses, has efficient message and time complexities, and lower overhead and superior timeliness properties than past thread scheduling algorithms. Our experimental studies validate our theoretical results, and illustrate the algorithms effectiveness.

On Scalable Synchronization for Distributed Embedded Real-Time Systems

We consider the problem of programming distributed embedded real-time systems with distributed dependencies. We show that the de facto standard of using locks and condition variables in conjunction with threads can have significant overhead and semantic difficulty and suggest alternative programming abstractions to alleviate these problems. We also discuss several alternatives for implementing these programming abstractions and discuss the algorithms and protocols needed.

On STM concurrency control for multicore embedded real-time software

We consider software transactional memory (STM) concurrency control in multicore embedded real-time software. We design an Earliest-Deadline-First (EDF) contention manager (CM) to augment STMs obstruction-free progress semantics. We establish the conditions under which STM/EDF-CM is competitive to lock-based and lock-free synchronization. Our experimental results reveal that STM/EDF-CM outperforms highly concurrent, fine-grain lock-based complex data structures, for which lock-free is often not viable.

Scheduling Dependent Distributable Real-Time Threads in Dynamic Networked Embedded Systems

We consider scheduling distributable real-time threads with dependencies (e.g, due to synchronization) in partially synchronous systems in the presence of node failure. We present a distributed real-time scheduling algorithm called DQBUA. The algorithm uses quorum systems to coordinate nodes’ activities when constructing a global schedule. DBQUA detects and resolves distributed deadlock in a timely manner and allows threads to access resources in order of their potential utility to the system. Our main contribution is handling resource dependencies using a distributed scheduling algorithm.

MCSAuth: A New Authentication Mechanism for Cloud Systems

Cloud Computing has attracted a lot of attention in both academia and industry lately. With its focus on scaling, collaboration, agility, availability and cost reduction, cloud computing offers a compelling alternative to in-house IT solutions. However, by “outsourcing” the computing infrastructure, it introduces a number of security issues. Specifically, since cloud computing is a shared computing platform, it needs to provide strong mechanisms for authenticating it’s users and ensuring that no confidential information stored on the cloud is compromised. This paper addresses a number of vulnerabilities in existing authentication mechanisms and proposes enhancements to mitigate these vulnerabilities. In addition, the paper studies how these enhancements affect performance. The results show that the existing vulnerabilities can be overcome by the proposed mechanisms. However, this results, in the worst case, in a six-fold increase in execution time. This can be considered relatively small when security is prime important.

Implementing distributable real-time threads in the Linux kernel: programming interface and scheduling support

We present an implementation of Real-Time CORBAs distributable threads (DTs) as a first-class, end-to-end realtime programming and scheduling abstraction in the Linux kernel. We use Ingo Molnars PREEMPT_RT kernel patch, which enables nearly complete kernel pre-emption, and add local (real-time) scheduling support to the Linux kernel, atop which we build DT scheduling support. We implement DTs using Linuxs threading capabilities. Our implementation of a suite of independent and collaborative DT schedulers confirm the effectiveness of our implementation.