Mobile Computing: State-of-the-Art and Future Trends
Lecturer:
Prof. Cristian Borcea
Cristian Borcea is an associate professor in the Department of Computer Science at the New Jersey Institute of Technology, USA. Cristian received his Ph.D. in Computer Science from Rutgers University, USA in 2004. Broadly, his research is on mobile computing, at the crossroad between systems and networking. More specifically, he focuses on designing, implementing, and evaluating programming models, middleware, network protocols, and security mechanisms for collections of ubiquitous wireless systems. His most recent projects are on mobile social computing and vehicular networks. Cristian’s research in the past few years has been funded by 6 National Science Foundation grants. He constantly publishes in top international journals and conferences. He also served on over 40 International Conference Technical Program Committees.
Place:
National Institute of Informatics, 20FL, Room 2010 (Sep.20: Room 2001)
This lecture presents an overview of mobile computing from its beginnings in the early 1990’s to these days. Specifically, it covers disconnected operation in mobile environments, mobile IP, wireless TCP, routing in mobile ad hoc networks, delay tolerant networking, power-aware mobile computing, mobile sensor networks, localization systems, and context-aware mobile computing. In this context, we will see how the challenges of mobile computing have evolved over the years, with some being solved, others becoming irrelevant, while new one appeared. The lecture ends with a brief presentation of the remaining lectures in the series.
This lecture has two parts. The first presents frameworks that leverage servers or cloud infrastructure to provide mobility support for users under Internet: (i) the Internet Suspend/Resume mechanism which layers a virtual machine (VM) on distributed storage, lets the VM encapsulate execution and user customization state, and makes the VM available to any machine the user wants to work on; (ii) systems for application decomposition and offloading that speed up the execution time and save battery power on mobile devices; and (iii) VM-based cloudlets, clusters located nearby mobile users, that provide cloud-like support for latency-sensitive mobile applications running on resource-constrained mobile devices. The second part of the lecture looks at mobility support in clean-slate network architectures, such as MobilityFirst or the GENI testbed, and discusses future networking challenges in this context.
This lecture will start with an introduction of mobile social applications, and then cover in detail several middleware platforms that can be used to simplify the programming and optimize the execution of such applications. We will compare centralized and decentralized middleware platforms and analyze the trade-offs between achieving a global view of the social state and protecting the user privacy. The second part of the lecture will discuss how mobile devices can be used to capture and learn human social behavior, and, in this context, present algorithms for community detection based on location or co-location traces collected from mobile users. The lecture concludes with a few examples of socially-aware protocols that use social information to improve response latency for mobile users.
This lecture covers mobile people-centric sensing, which can be a scalable and cost-effective alternative to deploying static wireless sensor networks for dense sensing coverage across large areas. The typical mobile sensors are smart phones and vehicular systems. For example, smart phones already have location, acceleration, audio, and video sensing capabilities, and in the near future are envisioned to include additional sensors, such as health and pollution monitoring sensors. The lecture focus will be mostly on application-specific systems for people-centric sensing. Examples of such systems include finding available parking spaces in the cities, finding potholes on the roads, and user/group activity recognition on smart phones. The lecture will conclude with a discussion of the challenges that face people-centric sensing, such as data reliability, user anonymity, and incentives.
This lecture shows how mobile ad hoc networks (MANETs) can be leveraged to provide distributed sensing and actuation in environments where Internet-based solutions do not work due to lack of connectivity, cost, or scalability. The main question in this context is how to program distributed MANET applications given the scale, heterogeneity, and volatility of these networks. We will present three programming models and their associated middeware platforms designed specifically to address these challenges: (1) Migratory Services, a context-aware client-service programming model; (2) Spatial Programming, a location-based imperative programming model; and (3) Contory, a SQL-like declarative programming model. All these models are implemented on top of the Smart Messages distributed computing platform, which provides naming, routing, and execution migration.
This lecture will cover applications, network architectures, and network protocols for vehicular computing. Examples of applications include automatic cab booking, monitoring the traffic ahead on the road, and dynamic traffic guidance. We will present 4 types of vehicular architectures: centralized, decentralized over ad hoc/opportunistic networks, decentralized over peer-to-peer networks, and hybrid (e.g., combination of servers based and ad hoc based techniques). The second part of the lecture will discuss routing and forwarding protocols for vehicular ad hoc networks, with a focus on protocols that leverage the road network and real-time vehicular traffic information to optimize data delivery.
This lecture will look at privacy and security aspects in people-centric sensing and ad hoc networks. We will first cover methods and architectures for maintaining privacy/anonymity of personal data, and especially location. Then, the lecture will discuss a complementary problem, which emphasizes the trade-off between privacy and trust, namely user location authentication for location-based services. The final part of the lecture covers two methods for creating trusted ad hoc networks at different network layers using software attestation and hardware support in the form of the Trusted Platform Module.
