Mohamed El-Darieby

Co-Scheduling Computational and Networking Resources in E-Science Optical Grids

With e-science applications becoming more and more data-intensive, data is generally generated and stored at different locations and can be divided into independent subsets to be analyzed distributed at many compute locations across an optical grid. It is required to achieve an optimal utilization of optical grid resources. This is generally achieved by minimizing application completion time, which is calculated as the sum of times spent for data transmission and analysis. We propose a Genetic Algorithm (GA) based approach that co-schedules computing and networking resources to achieve this objective. The proposed approach defines a schedule of when to transfer what data subsets to which sites at what times in order to minimize data processing time as well as defining the routes to be used for transferring data subsets to minimize data transfer times. Simulation results show the advantages of the proposed approach in both minimizing the maximum application completion time and reducing the overall genetic algorithm execution time.

Multidomain hierarchical resource allocation for grid applications

Geographically distributed applications in grid computing environments are becoming more and more resource intensive. Many applications require the collaboration between different domains, may be independently administrated domains, to exchange data and share computing and storage resources. This collaboration should be done in a way that maintains the privacy of each participant domain. This calls for new architectures and approaches to deal with such multidomain environments. We propose a hierarchical-based architecture as well as multidomain hierarchical resource allocation approach. The resource allocation is performed in a distributed way among different domains such that each participant domain keeps its internal topology and private data hidden while sharing abstracted information with other domains. Both computing and networking resources are jointly scheduled while optimizing the application completion time taking into account data transfer delays. Simulation results show the scalability and feasibility of the proposed approach.

Towards Network-Aware Divisible Load Theory for Optical Grids

Most grid applications require the processing of large amounts of data stored at different locations across the network which makes optical grid infrastructures optimal for such applications. The increase in intensity of data- and communications of these applications calls for new mechanisms and theories on how to optimally allocate optical grid resources. Typical Divisible Load Theory (DLT) makes optimal allocation of computational resources. In this paper, we introduce Network Aware Divisible Load Algorithm (NADLA) that extends DLT to optimally allocate both the computational and networking resources of an optical grid. We assume a data- and communications-intensive application where data is stored at different sites across the optical grid and can be divided into independent subsets to be processed in parallel at different sites. The algorithm defines an optimal data transfer schedule that defines when to transfer what data subsets to which sites across the optical grid in order to minimize the overall application completion time. It consists of two phases. In the first phase, NADLA provide load distribution that minimizes application completion time taking into account network connectivity and computational and networking resources availability. Site connectivity is estimated by considering the bandwidth and the expected free time of all the links connecting this site to other sites. In the second phase, a simple greedy algorithm is used to allocate computational and networking resources. Extensive Simulations are conducted to examine the performance of the proposed algorithm for different application types and sizes, and different optical grid topologies. Simulation results show the advantages of the proposed algorithm over the traditional DLT approach for the category of applications considered.

Vehicular Mobility Management Schemes for Balancing Loads among WMN Access Points

1 Abstract—Wireless mesh networks (WMN) [6] [7] are poised to be a cost-effective platform for many municipal applications in public safety, business and entertainment. In this paper we study WMN-based content delivery within Intelligent Transportation Systems (ITS) applications. An example ITS application would be to deliver content used for vehicle route guidance in emergency evacuation situations. We study and present three different vehicular mobility management schemes and compare these to the standard IEEE 802.11 mobility management scheme. We evaluate the benefits of the proposed scheme using OMNET++ simulator. Our results show that using the proposed schemes reduces handoff latencies and improves the overall network throughput at lower and higher vehicle speeds. In addition, the proposed schemes result in even distribution of the load of delivering ITS content to vehicles.

Scheduling big data applications within advance reservation framework in optical grids

Providing QoS for big data applications requires a way to reserve computing and networking resources in advance. Within advance reservation framework, a multi-domain scheduling process is carried out in a top down hierarchical way across multiple hierarchical levels. This ensures that each domain executes intra-domain scheduling algorithm to co-schedule its own computing and networking resources while coordinating the scheduling at the inter-domain level. Within this process, we introduce two algorithms: iterative scheduling algorithm and K-shortest paths algorithm. We conducted a comprehensive performance evaluation study considering several metrics that reflect both grid system and grid user goals. The results demonstrated the advantages of the proposed scheduling process. Moreover, the results highlight the importance of using the iterative scheduling and K-shortest paths algorithms especially for data intensive applications.

Resource information sharing in multi-domain optical grids

Providing interoperability in multi-domain optical grids requires sharing resource information among different domains. Sharing full resource information may not be appropriate due to security and business confidentiality reasons. We propose resource information sharing approach based on information aggregation and abstraction for both computing and networking resources. Information aggregation and abstraction affects the overall accuracy of information representation. This calls for more accurate information abstraction techniques. We propose three abstraction techniques: direct formulas, historical data and real-time measurements. The performance of the proposed information sharing approach and the abstraction techniques is evaluated by studying their impact on scheduling divisible load grid applications. Simulation results highlight the advantages of the proposed approach specially for data intensive applications. By using the proposed abstraction techniques, the same performance could be achieved as using large volumes of exact information.

Scalable and location-aware its content management in vehicular environments

ITS solutions require distributing content to vehicles traveling on streets and highways. This is becoming a more natural evolution in response to the pervasiveness of wireless devices in vehicles and urban areas. We propose a hierarchical architecture for ITS content delivery. The architecture utilizes the concepts and mechanisms of service-orientation to address the stringent requirements of vehicular requirements. The architecture also efficiently organizes services based on their location relevancy. We define algorithms and message structure used within the architecture to manage content services and efficiently delivery them to vehicles. Those algorithms include content service addition, updating, deletion, discovery, consumption and maintenance across different infrastructure nodes. We evaluate the benefits of the proposed scheme using OMNET++ simulator. Our results show that the architecture efficiently delivers content to vehicle traveling at moderate speeds as well as separated by average separation distances. Outside of these ranges, and due to extensive workloads of service delivery requests or requests to maintain delivery session, the communications overhead of the architecture increases decreasing its efficiency.

PCE-based hierarchical segment restoration

Providing network QoS involves, among other things, ensuring network survivability in spite of network faults. Fault recovery mechanisms should reduce recovery time, especially for real-time and mission-critical applications while guaranteeing QoS requirements, in terms of bandwidth and delay constraints and maximizing network resources utilization. In this paper, we propose a scalable recovery mechanism based on hierarchical networks. The proposed mechanism is based on inter-domain segmental restoration and is performed by a recovery module (RM) introduced for each domain of the hierarchy. The RM cooperates with Path Computation Element (PCE) to perform recovery while maintaining QoS. Segmental restoration ensures faster recovery time by trying to recover failed paths as close as possible to where the fault occurred. The recovery mechanism aggregates fault notification messages to reduce the size of the signaling storm. In addition, the recovery mechanism ranks failed paths to reduce recovery time for high-priority traffic. We present simulation results conducted for different network sizes and hierarchy structures. Two metrics were considered: recovery time and signaling storm size. A significant decrease in the recovery time with increasing number of hierarchical levels for the same network size is observed. The larger the number of hierarchy levels, the smaller the number of network nodes in each domain and, generally, the faster the routing computations and routing tables search times. In addition, the recovery mechanism results in reducing recovery time for high priority traffic by nearly 90% over that of lower priority traffic. However, increasing the number of hierarchical levels results in a linear increase in signaling storm size.

Estimation Vehicular Waiting Time at Traffic Build-Up Queues

Due to the high growth of social economic activities and the increased need for mobility in recent days, transportation problems like congestion, accidents, and pollution have been increased. However, improving the reliability of delay estimates and real-time dissemination of information remains a challenge. An advanced border-crossing system corresponding to the changes of cross-border circumstances becomes an urgent matter. An automated system for queue end monitoring has been proposed using image processing based transformed domain and empirical mode decomposition (EMD) feature extraction systems. The performance of feedforward backpropagation algorithm artificial neural networks (ANNs) was evaluated and tested, based on a selected set of features. The experimental results showed that the use of discrete wavelet transform (DWT) based Daubechies with decomposition of level 2 has accomplished the target with a processing time 2 sec and 3 epochs of training network only with best validation performance of (2.1053e-007) for vehicle recognition. Also the use of EMD as a feature extractor has accomplished the target of vehicle recognition with a best validation performance of (about 3.42e-09) and a processing time of 1 sec at epoch 3 of training network only with a minimal percentage of error for the recognition of each vehicle in the appropriate queue with the aid of the new concept of road side unit (RSU).

Mobile service-oriented content delivery in wireless mesh networks

Wireless mesh networks (WMN) is poised to be a cost-effective platform for many municipal applications in public safety, business and entertainment. In this paper we present WMN-based platform for content delivery within Intelligent Transportation Systems (ITS) applications. An example ITS application would be to deliver content used for vehicle route guidance in emergency evacuation situations. The characteristics of ITS applications of higher vehicle speeds and of the limited coverage of WMN (wi-fi) mesh router causes more frequent handoffs which complicates content delivery over wireless mesh networks. Frequent handoffs mandate smaller handoff delay. Using standard IEEE handoffs may cause unacceptable interruption in content delivery. We propose a service-oriented mobility management protocol (SMMP) for ITS content delivery. Within the protocol, content is considered as services described by XML metadata files. The protocol takes advantage of a hierarchical organization of WMN routers to reduce the handoff delay. The quasi-stationary nature of WMN mesh routers enables the detection of the sequence of routers that a vehicle will be in connection with. SMMP provides to the traveling vehicles cached MAC addresses of WMN mesh routers to communicate with. We evaluate the benefits of the proposed protocol and compared it to the traditional solutions using OMNET++ simulator. Our results show that using SMMP reduces handoff latencies and improves the overall network throughput at lower and higher vehicle speeds.