Jui-Ting Weng

Vehicular testbeds — Validating models and protocols before large scale deployment

Vehicular communications are becoming a reality due to the investments by stakeholders like car manufacturers and Public Transport Authorities. The building blocks of the ”Vehicle Grid” (radios, Access Points, spectrum, standards, etc.) are coming into place enabling a broad gamut of applications that range from navigation safety, intelligent transport, entertainment and urban sensing. Vehicular protocols and applications, however, must be carefully tested before deployment in the urban grid and introduction to the users. This validation must be carried out progressively in simulation, emulation and small scale testbed environments. In this paper we discuss the important role of the vehicular testbed in validating models and protocols before deployment in large scale scenarios. We illustrate the concept using two case studies that were carried out in the UCLA open vehicular testbed.

Recent advances and challenges in human-centric multimedia mobile cloud computing

The increasing usage of cloud computing, along with the proliferation of mobile devices and the demand for multimedia services, are changing the life style of users and creating new opportunities to providers and clients. Multimedia data will account for up to 90% of all Internet traffic in a few years, where most of the content will be created, shared, and accessed by mobile smartphones/tablets (carried by humans or placed in vehicles). However, this novel mobile multimedia era imposes new challenges for the networks, content, terminals, and humans, and must overcome problems associated, for instance, with high congestion, low scalability, fast battery consumption, and poor user experience. This paper discusses recent advances and challenges in human-centric mobile multimedia cloud computing approaches. On the one hand, Internet cloud will ubiquitously enrich multimedia mobile environments with more advanced and powerful features, including high processing and memory, scalability, availability, and adaptability. On the other hand, mobile devices will efficiently cooperate with each other to form mobile clouds that offload the Internet clouds from tasks that the latter cannot perform in a timely or efficient manner, including video and resource sharing. Both Internet and mobile clouds will be efficiently used to adapt/optimize multimedia flows to a single user or a group of users according to the current network conditions, context-awareness, content characteristics, device capabilities, and human experience.

Vehicular Testbeds - Model Validation before Large Scale Deployment

Vehicular communications are becoming a reality due to the investments by stakeholders like car manufacturers and Public Transport Authorities. The building blocks of the ”Vehicle Grid” (radios, Access Points, spectrum, standards, etc.) are coming into place enabling a broad gamut of applications that range from navigation safety, intelligent transport, entertainment and urban sensing. Vehicular protocols and applications, however, must be carefully tested before deployment in the urban grid and introduction to the users. This validation must be carried out progressively in simulation, emulation and small scale testbed environments. In this paper we discuss the important role of the vehicular testbed in validating models and protocols before deployment in large scale scenarios. We illustrate the concept using two case studies that were carried out in the UCLA open vehicular testbed.

Mobile social network based trust authentication

With the spread of mobile phones, through Wi-Fi and 3G communications, users have increased opportunities to retrieve messages and files posted by others while on the move, e.g. walking in the Mall, driving the car, etc. For safety, messages are often encrypted and the source is authenticated via PKI, preventing repudiation. Yet, at times the mobile may issue a message that is not accurate either innocently or maliciously. For example, the driver may report rocks on the pavement when there are crossing animals. Or he may maliciously report heavy traffic ahead to get other vehicles out of his way. The credibility of mobile report can be judged using a social network where the member is ranked by his friends. In this paper we review the construction of a trust graph in a social network like Facebook. We review popular ways of establishing the credibility of mobile user reports applying “transitivity” on single paths computed on the trust graph. We then argue for the importance of exploiting multiple paths in order to detect more subtle attacks not revealed by single paths. The methods are illustrated on artificial social trust networks (offering easily tunable parameters) and are then validated on real social network data.

Running consistent, parallel experiments in vehicular environment

The comparison of different applications and protocols in vehicular ad hoc networks (VANETs) is a very difficult task. This is mainly due to the nature of VANET, for which it is virtually impossible to replicate the same exact environment in two separate experiments. This paper introduces a flexible experiment structure that will allow different protocols to be compared in the same mobility pattern and channel conditions in VANET. A virtualized environment is setup on each node where the combination of Xen and Gentoo software is used to create multiple virtual guests. All virtual guests on the same machine use different protocols and share the wireless interface. We compare the widely recognized ad-hoc routing protocols AODV and OLSR to demonstrate the novelty of our system. Experiment results show that even with arbitrary external interference and mobility pattern, meaningful comparisons between different protocols can be carried out.

Surveillance service on the open mobile cloud

Mobile Vehicular Clouds are emerging as systems architectures suitable to provide services to both drivers and external customers. The paper describes a Pics-on-Wheel architecture and prototype implementation based on the idea of open mobile cloud. A simulation study based on San Francisco taxi cab traces shows that the Pics-on-Wheel service yields latencies in the order of 5 minutes in well frequented areas even with a modest number of vehicles. The results also show that by introducing incentives for active driver participation, even minimal detours noticeably reduce latencies.