Chalermek Intanagonwiwat

Desynchronization with an artificial force field for wireless networks

Desynchronization is useful for scheduling nodes to perform tasks at different time. This property is desirable for resource sharing, TDMA scheduling, and collision avoiding. Inspired by robotic circular formation, we propose DWARF (Desynchronization With an ARtificial Force field), a novel technique for desynchronization in wireless networks. Each neighboring node has artificial forces to repel other nodes to perform tasks at different time phases. Nodes with closer time phases have stronger forces to repel each other in the time domain. Each node adjusts its time phase proportionally to its received forces. Once the received forces are balanced, nodes are desynchronized. We evaluate our implementation of DWARF on TOSSIM, a simulator for wireless sensor networks. The simulation results indicate that DWARF incurs significantly lower desynchronization error and scales much better than existing approaches.

Declarative resource naming for macroprogramming wireless networks of embedded systems

Programming Wireless Networks of Embedded Systems (WNES) is notoriously difficult and tedious. To simplify WNES programming, we propose Declarative Resource Naming (DRN) to program WNES as a whole (i.e., macroprogramming) instead of several networked entities. DRN allows for a set of resources to be declaratively described by their run-time properties, and for this set to be mapped to a variable. Using DRN, resource access is simplified to only variable access that is completely network-transparent. DRN provides both sequential and parallel accesses to the desired set. Parallel, or group, access reduces the total access time and energy consumption because it enables in-network processing. Additionally, we can associate each set with tuning parameters (e.g., timeout, energy budget) to bound access time or to tune resource consumption.

V-DESYNC: Desynchronization for Beacon Broadcasting on Vehicular Networks

Several prospective applications on vehicular networks have been defined. Most applications rely on beaconing mechanisms to broadcast the presence and updated information of a vehicle to surrounding neighbors. However, due to the broadcasting nature, no acknowledgement mechanism is provided. Therefore, vehicles do not perceive beacon collision if two or more vehicles simultaneously broadcast the beacons. Consequently, vehicles miss updated information from their neighbors. In this paper, we propose V-DESYNC, an algorithm that distributively desynchronizes vehicles to broadcast beacons at different times based on only timing information. V-DESYNC is designed to avoid the beacon collision and tolerate the highly dynamic behavior of vehicular networks. Our evaluation results indicate that V-DESYNC can significantly reduce the number of beacon collisions without decreasing the beaconing rate on vehicular networks.

Energy-Efficient Gradient Time Synchronization for Wireless Sensor Networks

Wireless sensor network (WSN) applications usually demand a time-synchronization protocol for node coordination and data interpretation. In this paper, we propose an Energy-Efficient Gradient Time Synchronization Protocol (EGTSP) for Wireless Sensor Networks. In contrast to FTSP, a state-of-the-art synchronization protocol for WSNs, EGTSP is a completely localized algorithm that achieves a global time consensus and gradient time property using effective drift compensation and incremental averaging estimation. In contrast with GTSP, a gradient-based fixed-rated time synchronization protocol, our protocol provides adaptive beaconing for applications to optimize energy savings by selecting appropriate message-broadcast periods. The protocol is implemented and evaluated on multi-hop networks that consist of Telosb motes running TinyOS. The experimental results indicate that our protocol achieves a network-wide global notion of time, attains small synchronization errors, and utilizes energy efficiently.

A System for Using Wireless Sensor Networks as Globally Deductive Databases

Several research efforts have abstracted wireless sensor networks as relational databases whose data can be easily queried by users. However, none of these works includes recursive query mechanisms and globally logic reasoning techniques necessary for powerful uses as in deductive databases. In this paper, we propose an approach for abstracting wireless sensor networks as deductive databases that consist of rules, predicates, and logic-based queries. Our approach enables globally logic reasoning and data relating on a query that can be recursive or non-recursive. Additionally, the approach is more convenient to users because the users no longer need to be concerned about local behaviors of each sensor node. We also present LogicQ, an underlying system for disseminating and processing logic-based queries as well as collecting data in an energy-efficient manner. LogicQ uses a data filtering and suppressing technique that directly corresponds to an injected logic-based query for energy savings. Our performance analysis indicates that LogicQ can handle logic-based queries efficiently in terms of completeness and soundness while minimizing the energy consumption.

Logic Macroprogramming for Wireless Sensor Networks

It is notoriously difficult and tedious to program wireless sensor networks (WSNs). To simplify WSN programming, we propose Sense2P, a logic macroprogramming system for abstracting, programming, and using WSNs as globally deductive databases. Unlike macroprograms in previous works, our logic macroprograms can be described declaratively and imperatively. In Sense2P, logic macroprogrammers can easily express a recursive program or query that is unsupported in existing database abstractions for WSNs. We have evaluated Sense2P analytically and experimentally. Our evaluation result indicates that Sense2P successfully realizes the logic macroprogramming concept while consuming minimal energy as well as maintaining completeness and soundness of the answers.

A Set Cover-Based Density Control Algorithm for Sensing Coverage Problems in Wireless Sensor Networks

Wireless sensor networks consist of a large number of sensor nodes with limited power and resource. To prolong network lifetime, the energy consumption must be somehow reduced. In this paper, we propose a localized density control algorithm for energy savings. The goals are to maintain a minimal number of active sensor nodes and to reduce radio-traffic intensity while conserving the sensing coverage of the network. Our localized algorithm is based on a greedy solution of a weighted set-cover problem. Each node locally computes whether to sleep or to stay active. Given that the local decision might worsen the sensing coverage, we also introduce a voting scheme for selecting active nodes to assure that a node can sleep if and only if its sensing area is completely covered by its active neighbors. We have implemented our localized algorithm and voting scheme on Tiny OS and evaluated on TOSSIM. The result indicates that our algorithm is efficient and viable for practical use.

DNH-SaW: The Different Neighbor-History Spray and wait routing scheme for Delay Tolerant Networks

In Delay Tolerant Networks (DTNs), network partitioning is likely to happen due to sparse and mobile nodes with limited transmission ranges. The network partitioning leads to the unavailability of fully connected paths from sources to destinations. To deal with this problem, many routing schemes are proposed such as Epidemic routing and Spray and Wait routing. These schemes obtain high delivery rate but incur high overhead. In this paper, we propose a Different Neighbor-History Spray and Wait (DNH-SaW) routing scheme. DNH-SaW calculates the number of data copies to forward based on the neighbor-history of each receiver. Our simulation result shows that DNH-SaW can reduce overhead up to 27% and increases delivery utility up to 21.68% while achieving comparable delivery probability with previous works.

An Orthodontics-inspired desynchronization algorithm for wireless sensor networks

This paper proposes an Orthodontics-inspired desynchronization algorithm for scheduling wireless sensor nodes to avoid conflicts on resource sharing by not accessing the resource at the same time. Applications of desynchronization include TDMA scheduling, wake-sleep scheduling, and collision avoiding. Although existing desynchronization approaches perform reasonably well, their errors highly fluctuate. This high fluctuation implies overshooting and results in high errors even after convergence. For smoother convergence and lower error, we design a mechanism analogous to an orthodontic teeth-alignment technique. In Orthodontics, an orthodontist prevents the already-correct-positioning teeth from moving by tying them with a power chain. We apply the similar concept to the existing desynchronization method to prevent nodes with correct phases from adjusting their time phases. We evaluate our implementation of DESYNC-ORT on TOSSIM, a simulator for wireless sensor networks. The simulation result indicates that our method significantly helps the existing desynchronization algorithm to converge smoothly with lower error (reducing approximately 20–60% of errors caused by existing approaches).

Bittorrent peer identification based on behaviors of a choke algorithm

Bittorrent is currently one of the most popular peer-to-peer (P2P) file sharing protocols. However, it incurs such excessive amount of traffic that it may adversely affect users of legacy internet applications. To limit this adverse impact, an efficient methodology for bittorrent identification may be needed. In this paper, we propose a novel approach to identify local bittorrent peers. Our approach is based on behaviors of the choke algorithm, a main algorithm used in bittorrent. An advantage of our approach is that we can identify bittorrent peers without examining the packet pay-load. Therefore, our approach is free from privacy issues and still effective even though the packet payload is encrypted. Unlike previous works, we identify bittorrent hosts at the peer level instead of the flow level. Given that we use only information from network layer instead of transport layer, our work maintains fewer states and achieves robustness to changes in the transport layer. Furthermore, our work is effective even in restricted environments such as networks equipped with NAT devices or firewalls without modifications to existing network equipments. Our experimental result indicates that our approach can efficiently identify most of excessive bandwidth-consuming peers (i.e., peers transfer a large amount of data in our traces) with a low false-positive rate.