Technical Program

It's been quite a few years since the community has reached the conclusion that most published simulation studies of wireless ad hoc networks are not credible. The investigation of what undermines the faithfulness of these studies has been productive and has lead us to be cognizant of the various causes behind the problem. In this talk, I will discuss how we went from the understanding of these causes to the development of 'SWAN Tools', a system that protects simulation experimenters from their own mistakes. I will present details on the design and the implementation of this project, presenting directions for future developments.

In this talk we present a framework to perform credible simulation studies in the VANET research area. We present main modeling decisions and discuss open issues with emphasis on medium access and physical layer modeling. Further, we point out future perspectives and challenges of VANET simulations, not only on scenarios, but also on model validation and on scenario management.

In this case study we show how four topography-independent radio models (models not regarding individual obstacles in the operations area) can be "tuned" such that the outcomes of an event-driven network simulator come as close as possible to reference data obtained in a real-world radio network experiment. An evolutionary optimization process was applied to find near optimal parameter values four these models. We describe the optimization process, the error function that guides it and discuss the results.

Models of traffic demand are fundamental inputs to the design and engineering of data networks. This talk will address this requirement in the context of large-scale wireless infrastructures using real-measurement data. Our modeling effort focuses on capturing the demand variation in both the spatial and temporal domain in a way that scales well with the size of the wireless network. The network traffic dynamics are studied over two different monitoring periods at various levels of spatial aggregation, from individual buildings to the whole network.
Based on these models, we generated synthetic traffic for various spatio-temporal granularities and compared it with the measured (real-life) data. The comparison clearly illustrates the trade-off between model scalability and accuracy in capturing local-scale traffic dynamics. Our main contribution is a novel approach for traffic demand modeling in large wireless networks that features high flexibility in the exploitation of the spatial and temporal resolution available in data traces.
This talk will present these models, address the modeling tradeoffs, and analyze the performance of wireless hotspot APs with respect to throughput, goodput, delay and jitter per flow. The performance of these metrics will be computed under various traffic inputs, namely the measured data from a real-life network ("real-traffic") and synthetic traces based on various models. Τhe performance of these benchmarks using as input synthetic traffic based on popular/simple models differ substantially from the one acquired when the real-traffic is employed. Synthetic traces based on our models result in a performance very close to the one acquired when the real-traffic is used. Furthermore, the impact of various modeling parameters will be presented. Finally, the talk will also illustrate how these traces can be used in simulation and emulation testbeds for analyzing the performance of various wireless protocols.

Opportunistic networks are pervasive networks built exclusively by users' devices. With respect to legacy multi-hop ad hoc networks (MANETs), opportunistic networks are much more resilient to disconnections, long network partitions, and dynamic reconfigurations. While in MANETs nodes' mobility is an issue, in opportunistic networks mobility is an opportunity for communication, and therefore mobility models are a key component in opportunistic networks' research.
Improved (with respect to MANETs) users mobility models are rapidly becoming reference tools for studying opportunistic networks, thanks to their accuracy in matching statistical features observed in real traces. A very interesting approach is defining mobility patterns based on social relationships between users. This talk argues that, while being a fundamental building block, social relationships alone cannot be the only driver for the mobility process. Users movements are also driven by the attraction of physical locations (e.g., the house, the working place) on users. We analyse from this standpoint the Home-cell Community-based Mobility Model (HCMM), an extension of the social aware Community-based Mobility Model (CMM). HCMM retains the social-aware features of CMM and, in addition, permits to easily control the relationships between users and their preferred physical locations. There has been significant discussion in the research community on the fitting of key mobility metrics (e.g., inter-contact times, trip durations) against well-known distributions (such as exponential and power law), as the distribution of these metrics plays a major role in the performance of opportunistic network protocols. In the last part of the talk we present some preliminary results showing the characteristics of HCMM from this standpoint.

Mobile communication devices may be used for spreading multimedia data without support of an infrastructure. Such a scheme, where the data is carried by people walking around and relayed from device to device by means of short range radio, could potentially form a public content distribution system that spans vast urban areas. The transport mechanism is the flow of people and it can be studied but not engineered. The question addressed in this paper is how well pedestrian content distribution may work. We answer this question by modeling the mobility of people moving around in a city, constrained by a given topology. Our contributions are both the queuing analytic model that captures the flow of people and the results on the feasibility of pedestrian content distribution. Furthermore, we discuss possible approaches to extend the mobility model to capture the walking behavior that result from social interactions among pedestrians.

The field of delay tolerant networking is rich with protocols that exploit node mobility to overcome unpredictable or otherwise bad connectivity. The performance of many of these protocols is highly sensitive to the underlying mobility model which determines the nodes' movements, and the characteristics of these mobility models are not often studied or compared. With few exceptions, authors test their ideas using mobility models implemented on simulators written for the specific purpose of testing their protocols. We argue that it is better to unify these simulations to one highly capable simulator. We develop a suite of mobility models in OMNeT++ that specifically target delay tolerant networks. We also present a series of metrics that can be used to reason about mobility models independent of which communication protocols and data traffic patterns are in use. These metrics can be used to compare existing mobility models with future ones and also to provide insight into which characteristics of the mobility models affect which aspects of protocol performance. We implement a tool that derives these metrics from OMNeT++ simulations and implement several popular delay tolerant mobility models. Finally, we present the results of our analysis.