[show abstract][hide abstract] ABSTRACT: This paper presents PRESTO, a novel two-tier sensor data management architecture comprising proxies and sensors that cooperate with one another for acquiring data and processing queries. PRESTO proxies construct time-series models of observed trends in the sensor data and transmit the parameters of the model to sensors. Sensors check sensed data with model-predicted values and transmit only deviations from the predictions back to the proxy. Such a model-driven push approach is energy-efficient, while ensuring that anomalous data trends are never missed. In addition to supporting queries on current data, PRESTO also supports queries on historical data using interpolation and local archival at sensors. PRESTO can adapt model and system parameters to data and query dynamics to further extract energy savings. We have implemented PRESTO on a sensor testbed comprising Intel Stargates and Telos Motes. Our experiments show that in a temperature monitoring application, PRESTO yields one to two orders of magnitude reduction in energy requirements over on-demand, proactive or model-driven pull approaches. PRESTO also results in an order of magnitude reduction in query latency in a 1% duty-cycled five hop sensor network over a system that forwards all queries to remote sensor nodes.
IEEE/ACM Transactions on Networking 09/2009; · 2.01 Impact Factor
[show abstract][hide abstract] ABSTRACT: Structured overlays provide a basic substrate for many peer-to-peer systems. Existing structured overlays can be classi-fied into two categories, O(logN)-hop ones and O(1)-hop ones. The former are suitable for large and dynamic systems, while the latter are suitable for small or stable ones. However, for the lack of adaptivity, it is difficult for a peer-to-peer system designer to choose from these two kinds of overlays because the eventual size and churn rate are not easy to predict in the design phase. To address this problem, we propose a self-adaptive structured over-lay Tourist, which can adapt itself to the changing environment dynamically. On the one hand, Tourist can achieve 1-hop to 2-hop routing in most cases (e.g., in a 1,000,000-node system where nodes' average lifetime is only 1 hour). On the other hand, when the system size is extremely large or the nodes' churn rate is very high, Tourist can always guarantee O(logN)-hop routing for all the messages. Tourist nodes determine their routing table size autonomously: nodes with different capacities hold routing tables with different sizes. This makes Tourist sufficiently utilize all the nodes' allowable bandwidth to achieve as high routing efficiency as possible. Tourist also allows each node to adjust its routing table size dynamically, which is the essential reason for the self-adaptivity. Simulation results show that in a common 1,000,000-node system, Tourist can route all the messages within two hops and each node only pays no more than 1% bandwidth for its routing table maintenance. Only when the churn rate is very high (nodes' average lifetime being less than 1.5 minutes) Tourist would turn to an O(logN)-hop overlay where there are some mes-sages traversing more than two hops.
[show abstract][hide abstract] ABSTRACT: Nodes in peer-to-peer systems need to know the information about others to optimize neighbor selection, resource exchanging, replica placement, load balancing, query optimization, and other collaborative operations. However, how to collect this information effectively is still an open issue. In this paper, we propose a novel information collection protocol, PeerWindow, with which each node can collect a large amount of pointers to other nodes at a very low cost. Compared to existing protocols, PeerWindow is 1) efficient, the cost of collecting 1,000 pointers being less than 1 kbps in a common system environment, 2) heterogeneous, nodes with different capacities collecting different amounts of information, and 3) autonomic, nodes determining their bandwidth cost for node collection by themselves and adjusting it dynamically. PeerWindow can be used in many existing peer-to-peer systems and has tremendous potential for future expansions.
Parallel Processing, 2005. ICPP 2005. International Conference on; 07/2005
[show abstract][hide abstract] ABSTRACT: Granary is a public storage service on the Internet that has two distinguished goals in comparison with previous projects. First, it is object-oriented, and thereby supports attribute-level data query. Second, it is very flexible to the system environment, i.e., it can be deployed in a grid-like environment, a peer-to-peer-like environment, or even a compromised one. In this paper, we present Granary's architecture, as well as some of its significant components that are designed in adherence to these two goals, including the node-collection protocol PeerWindow, the routing infrastructure Tourist, and the object-index management algorithm PB-link Tree. An implementation of Granary is in development and intended to be deployed in a campus scale
E-Commerce Technology for Dynamic E-Business, 2004. IEEE International Conference on; 10/2004
[show abstract][hide abstract] ABSTRACT: The high heterogeneity of large-scale p2p system leads us to the philosophy that the size of a node's routing table and its updating cost should correspond to the node's capacity. With this philosophy, we design a novel structured overlay: SmartBoa. SmartBoa categorizes nodes into different levels according to their capacities. A node at level k has a routing table with k N 2 / entries (N is the system scale). An efficient non-redundant multicast algorithm is introduced to distribute nodes' changing reports, with which the routing table's updating cost is in proportion to its size. Node can change its level freely to adapt to the fluctuation of bandwidth. At the same cost as the Pastry-like overlay, SmartBoa maintains rather larger routing tables and has much higher routing efficiency. A low-bandwidth (64 kbps) node can maintain 10,000 routing entries at the cost of only 10 percent of its bandwidth. Without the high bandwidth requirement of one-hop overlay, SmartBoa is much more scalable.
Peer-to-Peer Systems III, Third International Workshop, IPTPS 2004, La Jolla, CA, USA, February 26-27, 2004, Revised Selected Papers; 01/2004
[show abstract][hide abstract] ABSTRACT: In research of overlay networks, simulator takes a very important role. However, popular simulators, such as ns and PlanetLab can't meet the scale and performance requirement of overlay research. Although it is necessary for overlay researchers to observe activities of million nodes, current simulators can not give such simulation result. To improve the overlay network research efficiency, we design and implement ONSP, a novel parallel overlay network simulation platform, which provides parallel discrete event simulation of overlay networks on high performance cluster. With this tool, we are able to build overlay network simulator in large scale easily and test it in short time. The test result proves that ONSP can well address the requirement of performance and scalability.
Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications, PDPTA '04, June 21-24, 2004, Las Vegas, Nevada, USA, Volume 3; 01/2004
[show abstract][hide abstract] ABSTRACT: Although peer-to-peer overlays have been elaborately studied, they still have redundant overheads. Pastry and Chord use leaf sets to connect those nodes contiguous in numerical space, but these pointers have no contribution to efficient routing. In this paper we argue that if the mapping between a message key and its root node is determined in a smart way, leaf sets can be removed, largely decreasing system overheads. We present a novel overlay algorithm derived from Pastry, in which nodeIds have different lengths and form a Huffman set, routing tables have no empty items and root node exactly pre- fix-matching its keys. By these means, our approach abolishes the leaf sets with no ill effect on routing performance or other properties. Experimental results show that our approach saved about 22~25% messages cost for maintenance from Pastry in an overlay of 10,000 peers.
Grid and Cooperative Computing, Second International Workshop, GCC 2003, Shanghai, China, December 7-10, 2003, Revised Papers, Part I; 01/2003
[show abstract][hide abstract] ABSTRACT: Teaching, if done well, can be the most exciting and rewarding part of an academic career. As a student, I have been fortunate to learn several effective teaching techniques from some great teachers, and have practiced these techniques as a teaching assistant and as a mentor of junior students. These teaching techniques include building an interactive atmosphere in the classroom, providing real-world problems and hands-on experience, and encouraging collaborations. • An interactive atmosphere in the classroom is one of the most important elements of successful teaching. Teachers who talk all the time without any interaction with the class make students lose interest and are less successful. Instead, when teaching a course, I will introduce problems and basic knowledge first, then direct students to solve the problems by asking questions and leading discussions. I have successfully practiced this technique in several lectures and talks. • Real-world problems and hands-on experience inspire students' interest, help them under-stand abstract material, and can be useful to them outside the class. I would like to design course projects with real-world problems on real systems, instead of simulated "classroom-oriented" sys-tems. Simulated systems are easy for students to handle, but can never give them the same experience as complex actual systems. • Collaboration between students can dramatically speed up the learning process, and there-fore should be encouraged. Few students can learn efficiently in a vacuum, and discussing hard problems with classmates can be a harmless shortcut. Group projects encourage collaborations between students, and can help them to build collaboration skills that are required in most engi-neering and research work today.