This paper presents a database containing 'ground truth' segmentations produced by humans for images of a wide variety of natural scenes. We define an error measure which quantifies the consistency between segmentations of differing granularities and find that different human segmentations of the same image are highly consistent. Use of this dataset is demonstrated in two applications: (1) evaluating the performance of segmentation algorithms and (2) measuring probability distributions associated with Gestalt grouping factors as well as statistics of image region properties.

/pdf/a-database-of-human-segmented-natural-images-and-its-2ve1vk356s.pdf

A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

https://deim.urv.cat/~alexandre.arenas/publicacions/pdf/review.pdf

Synchronization in complex networks

This survey covers rollback-recovery techniques that do not require special language constructs. In the first part of the survey we classify rollback-recovery protocols into checkpoint-based and log-based.Checkpoint-based protocols rely solely on checkpointing for system state restoration. Checkpointing can be coordinated, uncoordinated, or communication-induced. Log-based protocols combine checkpointing with logging of nondeterministic events, encoded in tuples called determinants. Depending on how determinants are logged, log-based protocols can be pessimistic, optimistic, or causal. Throughout the survey, we highlight the research issues that are at the core of rollback-recovery and present the solutions that currently address them. We also compare the performance of different rollback-recovery protocols with respect to a series of desirable properties and discuss the issues that arise in the practical implementations of these protocols.

/pdf/a-survey-of-rollback-recovery-protocols-in-message-passing-bnrpate93a.pdf

A survey of rollback-recovery protocols in message-passing systems

Parallel discrete event simulation (PDES), sometimes called distributed simulation, refers to the execution of a single discrete event simulation program on a parallel computer. PDES has attracted a considerable amount of interest in recent years. From a pragmatic standpoint, this interest arises from the fact that large simulations in engineering, computer science, economics, and military applications, to mention a few, consume enormous amounts of time on sequential machines. From an academic point of view, parallel simulation is interesting because it represents a problem domain that often contains substantial amounts of parallelism (e.g., see [59]), yet paradoxically, is surprisingly difficult to parallelize in practice. A sufficiently general solution to the PDES problem may lead to new insights in parallel computation as a whole. Historically, the irregular, data-dependent nature of PDES programs has identified it as an application where vectorization techniques using supercomputer hardware provide little benefit [14].A discrete event simulation model assumes the system being simulated only changes state at discrete points in simulated time. The simulation model jumps from one state to another upon the occurrence of an event. For example, a simulator of a store-and-forward communication network might include state variables to indicate the length of message queues, the status of communication links (busy or idle), etc. Typical events might include arrival of a message at some node in the network, forwarding a message to another network node, component failures, etc.We are especially concerned with the simulation of asynchronous systems where events are not synchronized by a global clock, but rather, occur at irregular time intervals. For these systems, few simulator events occur at any single point in simulated time; therefore parallelization techniques based on lock-step execution using a global simulation clock perform poorly or require assumptions in the timing model that may compromise the fidelity of the simulation. Concurrent execution of events at different points in simulated time is required, but as we shall soon see, this introduces interesting synchronization problems that are at the heart of the PDES problem.This article deals with the execution of a simulation program on a parallel computer by decomposing the simulation application into a set of concurrently executing processes. For completeness, we conclude this section by mentioning other approaches to exploiting parallelism in simulation problems.Comfort and Shepard et al. have proposed using dedicated functional units to implement specific sequential simulation functions, (e.g., event list manipulation and random number generation [20, 23, 47]). This method can provide only a limited amount of speedup, however. Zhang, Zeigler, and Concepcion use the hierarchical decomposition of the simulation model to allow an event consisting of several subevents to be processed concurrently [21, 98]. A third alternative is to execute independent, sequential simulation programs on different processors [11, 39]. This replicated trials approach is useful if the simulation is largely stochastic and one is performing long simulation runs to reduce variance, or if one is attempting to simulate a specific simulation problem across a large number of different parameter settings. However, one drawback with this approach is that each processor must contain sufficient memory to hold the entire simulation. Furthermore, this approach is less suitable in a design environment where results of one experiment are used to determine the experiment that should be performed next because one must wait for a sequential execution to be completed before results are obtained.

/pdf/parallel-discrete-event-simulation-m759whzicd.pdf

Parallel discrete event simulation

A number of library based parallel and sequential network simulators have been designed. The paper describes a library, called GloMoSim (Global Mobile system Simulator), for parallel simulation of wireless networks. GloMoSim has been designed to be extensible and composable: the communication protocol stack for wireless networks is divided into a set of layers, each with its own API. Models of protocols at one layer interact with those at a lower (or higher) layer only via these APIs. The modular implementation enables consistent comparison of multiple protocols at a given layer. The parallel implementation of GloMoSim can be executed using a variety of conservative synchronization protocols, which include the null message and conditional event algorithms. The paper describes the GloMoSim library, addresses a number of issues relevant to its parallelization, and presents a set of experimental results on the IBM 9076 SP, a distributed memory multicomputer. These experiments use models constructed from the library modules.

GloMoSim: a library for parallel simulation of large-scale wireless networks

Building a National Modeling and Simulation (M&S) Coalition

The authors discuss an approach to federated network simulations that eases the burdens on the simulation developer in creating space-parallel simulations. Previous approaches have had difficulties that arise from the need for global topology knowledge when forwarding simulated packets between federates. In all but the simplest cases, proper packet-forwarding decisions between federates requires routing tables of size O(mn), where m is the number of nodes modeled in a particular federate, and n is the total number of network nodes in the entire topology. Furthermore, the benefits of the well-known NIx-Vector routing approach cannot be fully achieved without global knowledge of the overall topology. The authors overcome these difficulties by using a topology partitioning methodology that uses ghost nodes. They show experimentally that the memory overhead associated with ghosts is minimal relative to the overall memory footprint of the simulation.

Using Ghosts for Global Topology Knowledge in Space-Parallel Distributed Network Simulations

This thesis focuses on two important problems related to DDDAS: interest management (data distribution) and prediction models. In order to reduce communication overhead, we propose a new interest management mechanism for mobile peer-to-peer systems. This approach involves dividing the entire space into cells and using an efficient sorting algorithm to sort the regions in each cell. A mobile landmarking scheme is introduced to implement this sort-based scheme in mobile peer-to-peer systems. The design does not require a centralized server, but rather, every peer can become a mobile landmark node to take a server-like role to sort and match the regions. Experimental results show that the scheme has better computational efficiency for both static and dynamic matching. 
In order to improve communication efficiency, we present a travel time prediction model based on boosting, an important machine learning technique, and combine boosting and neural network models to increase prediction accuracy. We also explore the relationship between the accuracy of travel time prediction and the frequency of traffic data collection with the long term goal of minimizing bandwidth consumption. Several different sets of experiments are used to evaluate the effectiveness of this model. The results show that the boosting neural network model outperforms other predictors.

/pdf/interest-management-scheme-and-prediction-model-in-3lnhmumx48.pdf

Interest management scheme and prediction model in intelligent transportation systems

Evaluating the effects of a change to one aspect of the modern urban environment requires knowledge about other parts and how they are interconnected. Thus, integrated simulations of many parts of the entire urban ecosystem are needed to create and evaluate the effects of policies. Many existing models target a single urban system - land use models, transportation models, water system models, etc. - however coupling them to create interconnected systems often requires great effort and resources. In this work we propose a common object model for integrated transportation and land use models (ITLUMs) using the High Level Architecture Standard (IEEE 1516) (IEEE 2010). This object model is the first step in a longer term effort to develop new simulation models that can be readily integrated through common data abstractions. Finally, the proposed object model is designed to be extendable to capture the elements from other models of the urban environment.

Toward a common object model for integrated transportation and land use models

We investigate whether speculative execution can be used in real-time computations. We explore the use of speculative execution in real-time systems in general and demonstrate the application of our methods in real-time discrete-event-simulation in particular. The rationale behind choosing the latter paradigm is that a well-known optimistic protocol called Time Warp has been successfully employed to parallelize irregular simulations. However, the use of such methods for real-time simulation remains largely unexplored.
We develop a parameterized model for optimistic real-time computations. The model accounts for real-time, and for overheads such as state saving, state restoration, and sending and transmitting positive and negative messages. Based on this model, we derive necessary and sufficient conditions for speculatively-executing real-time computations to keep their deadlines. Further, we show that optimistic simulators that never send incorrect messages conform to the necessary and sufficient conditions derived above, and thus provide sufficient predictability to allow schedulability analysis techniques in extant real-time systems to be applied.
Considering this issue of programmability vs. performance, we point out the problems that arise in using shared state in speculative computations, and investigate the performance of a software abstraction called Space-Time memory to address these problems. This abstraction allows any speculative protocol that relies on timestamped transactions to interact via shared memory rather than message-passing.
To avoid thrashing, fast and frequent memory reclamation can become critical in speculative real-time systems. We describe a scheme for reconfiguring garbage-collection of our data structures vis-a-vis the amount of memory to be used and the idle-time intervals 'currently available' in the system. The parameters of the reconfiguration are based on the hardware platform, and the amount of idle time available in the real-time system.
We demonstrate the feasibility of our ideas by developing a testbed for optimistic real-time computation on a 64-node Kendall Square Research machine. The performance of this prototype with respect to synthetic workloads and a robot-control application is encouraging.

Richard M. Fujimoto

Papers

Building a National Modeling and Simulation (M&S) Coalition

Using Ghosts for Global Topology Knowledge in Space-Parallel Distributed Network Simulations

Interest management scheme and prediction model in intelligent transportation systems

Toward a common object model for integrated transportation and land use models

Speculative execution in real-time systems