Geo-location is critically important in many mobile applications including mobile social network and u-health applications. The proliferation of WLAN infrastructures can be usefully utilized for indoor localization using WLAN fingerprints. Our work shows that identifying a room or a place is feasible as tens of accessible APs are accessbile in typical urban environments. We design and implement a cooperative indoor localization technique using Android-based mobile devices for mobile social network applications.

Another important research agenda in this area is to identify mobile devices by characterizing their analog signals. While authentication and security measures at higher layers can be compromised by malicous attackers, their analog signals are not easily subjective to random manipulation. This research could help to offer secure u-health applications. We are using a combination of USRP and GNU Radio, which is identified as one of Key Technologies to Watch in 2008 by Bristol Systems. Click here for more information on USRP / GNU Radio. Also, see a class website.

Mobile networks are easily vulnerable to stresses such as high traffic and high node speed as well as privacy and security attacks. As existing link and network layer solutions do not take these extreme conditions into consideration, it is critically important to know whether a mobile network is still a dependable subnet under such situations. The goal of this research is to seek novel methods to survive the stress while achieving a reasonable performance. We are exploring several measures including bandwidth subchanneling.

We analyze the computational complexity of communication functions using GNU Radio SDR software to take communication and computation in a unified framework. Several optimizations are possible; i.e., rate-distance tradeoff turns into rate-distance-energy tradeoff. Currently we're profiling several modulation and coding schemes by using Oprofile profiling tool.

This work is currently supported in part by National Science Foundation, Major Research Instrumentation (MRI) Program. The project is entitled "Performabiity in Mobile Wireless Networks."

We designed and experimented Multihop Opportunistic Transmission Protocol (MTOP) that significantly advances the communication efficiency in a mixture of devices with different bit rates. It is being integrated with bandwidth scaling and modulation diversity techniques to further improve the network performance. We also designed Path-centric Rate Adaptation (PRAM) algorithm, which is a cross-layer routing scheme. It is being tested using ns-2 simulation as well as Emulab emulation environment.

This work is currently supported in part by National Science Foundation, Network Technology and Systems (NeTS) Program. The project is entitled "Exploring Data Access in Internet-based Wireless Mobile Networks"

Inspired by the Smart Dust project started in 1998, researchers have shown enormous interests in wireless sensor networks (sensornets) because of their long-term potential in many interesting pervasive applications. However, one of the challenges for sensornets to become a prevailing technology in the next decade comes from the complexity of managing the raw data distributed in a sensornet and turning it into information for decision-making purposes. Two major sources of the complexity are resource constraints and uncertainties. While the former (e.g., severe energy constraints and limited bandwidth) has attracted a lot of attention from the research community, little work has been done on the complexity due to ontological and epistemic uncertainties. Ontological uncertainty can emerge from the lack of specification of what kind of entities could exist and epistemic uncertainty emerges due to inadequate representation of knowledge that is often incomplete, imprecise, fragmentary, and ambiguous. The main goal of the proposed research is to develop a conceptual framework that deals with those uncertainties based on classical rough set theory (Wiki page) and to reconsider sensornet architectures and algorithms in a way to support as well as exploit the framework.

For example, one can imagine a surveillance network for contaminant detection in water distribution system. A conventional detection problem asks for a tradeoff between the probability of detection and the false alarm rate because the consequence of a warning could be the stoppage of the infrastructure service which is quite a costly measure, the false alarm must be avoided as much as possible. However, a major challenge is that there are so many contaminants that a large array of (contaminant-specific) sensors might be required, while still leaving the system vulnerable to contaminants for which no effective sensor was available or not employed. This ontological uncertainty can be approached by monitoring more general water parameters to identify a signature of “normal” conditions, and flag anomalies. This would require a more significant sensornet component, and would involve a more substantive “multi-parameter filtering” problem - to estimate the state of the system and derive a conditional probability that the condition is normal or anomalous. Questions to how many sensors, where to sense, and how often to sense constitute epistemic uncertainty: the observation of coarser granularity offers less detail while the clumping of information into an aggregate form may prevent finer entities from being distinguished. While uncertainty in general affects a systems ability to perform with accuracy and precision, the impact of uncertainty in sensornets. We are also considering to apply the same idea in sensornet-based rehab applications.

Extrapolating from the phenomenal success of social network services, it is not unrealistic to anticipate that mobile social networking would play an important role in the near future. However, since social activity would be inherently transient due to user mobility rendering the corresponding interaction spontaneous, challenges are to discover interaction opportunities spontaneously and to support interactions while taking the limited resource of users’ mobile devices and the scarce wireless bandwidth into consideration. To address these challenges, We design and implement autonomic networking architecture called SINA. It exploits contexts such as application requirements, user preferences, the availability of services and resources as well as topology of nodes to find an interaction opportunity. A minimal set of resources are used because (i) a network instance is created only when an interaction opportunity is found and (ii) it is organized and maintained in an application-aware manner.

Wireless networks are increasingly popular as the last-mile solution for a ubiquitous communication infrastructure. This trend in combination with the growing interest in accessing vast amount of resources in Internet has driven the developments of hybrid wireless network architectures such as Internet-based mobile ad hoc networks (IMANETS) and Internet-based vehicular ad hoc networks (IVANETS). For instance, the backbone network of wireless mesh networks is typically architected as an IMANET. One of key features of these networks is multi-hopping, which extends the coverage of a wireless network without investing for an additional infrastructure. However, it is well-researched that these multi-hop wireless networks possess the scalability problem in the sense that per-client bandwidth is critically limited and decreasing as the network size grows. In order to improve the data accessibility, which refers to the capability of allowing mobile users to access desired data with high success rate, a prudent data caching scheme is required. In multi-hop wireless networks, a successful data caching scheme stores data not only in Internet gateways but also in individual client devices as they typically play as intermediate routers. Correspondingly, three key design issues are: which data to cache (limited storage space), how to invalidate cached data (invalidation message explosion), and how to implement and operate relaxed cache consistency model in IMANETS and IVANETS.

Therefore, this project aims at developing algorithms and communication protocols that allow efficient and correct data caching in Internet-based wireless mobile networks. In this proposal, we investigate the following four major areas: (i) Cache management scheme for IMANETS: We investigate the problem of information search and access, and propose cache management mechanisms including a cache admission control and a cache replacement policy that improve the overall information accessibility; (ii) Cache invalidation strategies for IMANET: For providing data consistency in an IMANET, we propose several push and pull-based cache invalidation schemes, the design of which is not straightforward due to the complications caused by multi-hop message relay, operation cost model, and uncertainty in message delivery; (iii) Cache invalidation scheme for IVANETS: To access the Internet service and information on the wheel, we consider the integration of VANET with wireless infrastructure. Under this environment, cache management (which data to cache) is less of a problem because the storage space in a vehicle is not critically limited. On the other hand, high host mobility complicates the invalidation process. We propose a location-based cache invalidation scheme to reduce the negative impact of mobility, the cost of broadcast, and the query latency; and (vi) Cache consistency strategies for IVANETS: Strong cache consistency is not easily achievable or is too costly to achieve in IVANETS due to fast-changing network topologies. We propose weak cache consistency strategies that make a tradeoff between cache consistency and bandwidth requirements for ensuring the consistency.