SensorNet Paper Reviews by rcs38
From Ece
- Return to the paper reading schedule or the Sensor Networks homepage.
Sensor Networks for Emergency Response: Challenges and Opportunities
K. Lorincz, D. Malan, T.R.F. Fulford-Jones, A. Nawoj, A. Clavel, V. Shnayder, G. Mainland, S. Moulton, and M. Welsh, Sensor Networks for Emergency Response: Challenges and Opportunities, In IEEE Pervasive Computing, Special Issue on Pervasive Computing for First Response, Oct-Dec 2004.
This paper focuses on the applications in which sensor networks could be and are currently used. Particular attention is devoted toward Emergency Response applications as well as general medical applications. The authors are currently developing a sensor node infrastructure named “Code Blue.” The criteria that a well developed infrastructure should meet are: discovery and naming, robust routing, prioritization of critical data, security, and tracking device locations. The current Code Blue system allows Mica2 motes to communicate with hand held PDAs. The Code Blue design team has successfully written software and firmware to allow a PDA to monitor patient’s heart rate, blood oxygen saturation, and heart electrical activity. The entire system is wireless and allows first response teams to assess a patients health before arriving to a hospital. One draw back to the current system is security. The team has yet to design a protocol that can handle secure data. If two emergency response teams need to work together, how will the sensors know which devices can be trusted with secure information.
Habitat Monitoring with Sensor Networks
R. Szewczyk, E. Osterweil, J. Polastre, M. Hamilton, A. Mainwaring, and D. Estrin, Application Driven Systems Research: Habitat Monitoring with Sensor Networks, Communications of the ACM, Special Issue on Sensor Networks. June 2004.
This paper discusses the needs and demands of a wireless habitat monitoring sensor network. The main focus is on communication and routing as well as power consumption. The main infrastructure consists of sensor nodes which transmit data to nearby sensor nodes which in turn propagate the information from one node to another until it reaches a data collection station. Various techniques are used to accomplish this daunting data transmission/collection process. Carrier-sense multiple access (CSMA) periodically listens to different channels. If activity is on the channel then the node will capture the packets it senses and retransmit them to other nodes. This technique is also referred to as low-power listening (LPL). Although different options exist for constructing network architectures, the options for low power applications are few. The main setback of the habitat monitoring network is power consumption. Wireless communication for sensor nodes requires a significant amount of energy and little volume to store it in, an area in which battery technology seems to lack. Nodes have yet to become maintenance free and require battery changes periodically.
Connecting the Physical World with Pervasive Networks
D. Estrin, D. Culler, K. Pister, and G. Sukhatme, Connecting the Physical World with Pervasive Networks , IEEE Pervasive Computing, pp. 59-69, January-March 2002.
This paper gives an overview of sensor networks and the limitations, restrictions, and demands of such networks. According to the paper, the technology of sensor networks is evolving rapidly. Physical devices are getting smaller each day and energy consumption is lowering as well. The many uses of ultra-small, ultra-low power wireless sensor networks is growing rapidly. These devices can be embedded in almost anything and exist completely unnoticeable to the substance, environment, or organism they monitor. It is predicted that in 10 years devices which can sense the environment, store information, and wirelessly send that information to other devices will exist in a volume no more than one cubic millimeter. While sensor networks are in use today, no network exists on such a small scale. Technology has a good ways to go before such advances will be realized.
Towards a Sensor Network Architecture: Lowering the Waistline
D. Culler, P. Dutta, C.T. Ee, R. Fonseca, J. Hui, P. Levis, J. Polastre, S. Shenker, I. Stoica, G. Tolle, and J. Zhao, Towards a sensor network architecture: Lowering the waistline. Proceedings of Hot Topics in Operating Systems (HotOS '05), 2005
This paper discusses the lack of a uniform network architecture for sensor networks. The author proposes a Sensor-net Protocol (SP) which provides an architecture for sending and receiving information, establishing connections, discovery, and naming. The goal of SP is to accomplish communications abilities for sensor networks in the same fashion that Internet Protocol (IP) accomplished for communication over the internet. This approach allows many different devices which were constructed for many different purposes to communicate with any other device, regardless of a devices purpose. SP is proposed to me a lower level protocol than IP enabling it to accomplish address free and name-based communication. The concepts in this paper are tantalizing but remain mere concepts. The idea has yet to be realized and only exists in a conceptual form.
MANTIS OS: Am Embedded Multithreaded Operating System for Wireless Micro Sensor Platforms
S. Bhatti, J. Carlson, H. Dai, J. Deng, J. Rose, A. Sheth, B. Shucker, C. Gruenwald, A. Torgerson, R. Han, "MANTIS OS: An Embedded Multithreaded Operating System for Wireless Micro Sensor Platforms, " ACM/Kluwer Mobile Networks & Applications (MONET), Special Issue on Wireless Sensor Networks, vol. 10, no. 4, August 2005
This paper presents a new approach in wireless sensor network operating systems. The difference between the proposed operating system (MANTIS) and existing operating systems is that MANTIS is based on time slots for processes, versus event triggers such as those utilized in Tiny OS. The driving force behind the development of MANTIS is the need to reduce power consumption in wireless sensor network nodes. MANTIS operates in a miniscule 500 bytes of code, allowing a large percentage of the small memory available on sensor nodes to be dedicated toward processes. The low power operation was achieved by invoking a sleep state for the microprocessor whenever the last active process finished its execution. One subject that the paper failed to address is that of security. If security must be implemented for data transmission over the network, then how much overhead will it add to the operating system? This would also have a directly negative effect on power consumption.
The Emergence of Networking Abstractions and Techniques in TinyOS
P. Levis, S. Madden, D. Gay, J. Polastre, R. Szewczyk, A. Woo, E. Brewer and D. Culler, The Emergence of Networking Abstractions and Techniques in TinyOS, USENIX/ACM Symposium on Networked Systems Design and Implementation (NSDI). 2004.
This paper examines different implementation techniques for TinyOS. These techniques are categorized into four different levels of abstraction. General abstractions are abstractions which are widely used and TinyOS provides both the mechanisms and policies to support them. Specialized abstractions are those that provide their own policy and simply use the mechanisms provided by TinyOS. In flux abstractions are those that are implemented as part of an application. They provide the mechanisms and policies to execute. Absent abstractions are those that appear frequently in text but are hard to come by in actual implementation. The paper concludes that since wireless sensor nodes are designed around power management, limited resources, and real-time constraints, no common census can be used as a universal abstraction method.
A Dynamic Operating System for Sensor Nodes
C.-C. Han, R.K. Rengaswamy, R. Shea, E. Kohler and M. Srivastava. A Dynamic Operating System for Sensor Networks, International Conference on Mobile Systems, Applications, and Services (MobiSys), 2005.
This paper presents an idea for a new breed of operating systems that it calls SOS. SOS is an operating system that allows dynamic resource allocation while maintaining an energy performance similar to that of TinyOS. One interesting advantage of SOS is that the code updates are significantly smaller than that of TinyOS. This subtle difference has dramatic effects in power conservation during the update process. However, since updates to the OS do not occur frequently, the overall effect of this power conserving feature is miniscule. Since the overall power consumption for SOS is comparable to that of TinyOS, yet SOS offers dynamic resource allocation and an easier updating process, SOS is a worthy consideration in the choice for sensor node operating systems.
Just-In-Time Sensor Networks
G. Yee, B. Shucker, J. Dunn, A Sheth, R Han, "Just-In-Time Sensor Networks", IEEE Workshop on Embedded Networked Sensors (EmNets) 2006, pp. 6-10.
This completely conceptual paper describes a process for quick and easy deployment of sensor nodes. The paper introduces the idea of a “cannon” which would program the device and also deploy the device. One application would be the analysis of the land after a mud slide. Sensor nodes would be quickly deployed with the use of the “cannon.” This first wave of deployment would be less dense and would collect general information about the area it is surveying. The information from these nodes would be used to choose locations to place a denser patch of second wave sensor nodes. Each subsequent wave of nodes delivers finer grain detail about the area. While these devices should be cost effective, easily deployable, and apparently reprogrammable, the authors give no method for actually constructing such a device. The paper outlines some issues that will need to be addressed, but does not actually address any of them.
Temporal Properties of Low Power Wireless Links: Modeling and Implications on Multi-Hop Routing
A. Cerpa, J.L. Wong, M. Potkonjak, and D. Estrin, Temporal Properties of Low Power Wireless Links: Modeling and Implications on Multi-hop Routing, ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc'05), May 2005.
PowerPoint Slides
On Calculating Power-Aware Connected Dominating Sets for Efficient Routing in Ad Hoc Wireless Sensor Networks
Jie Wu, Ming Gao and Ivan Stojmenovic, On calculating power-aware connected dominating sets for efficient routing in ad hoc wireless networks, submitted for publication, 2000.
This paper introduces a new algorithm for organizing ad-hoc wireless sensor nodes. Ad-hoc networks (in the realm of sensor nodes) tend to construct themselves in such a way that they naturally form groups of connected nodes. Within each group (set) of connected nodes, a few nodes exist that communicate with nodes in neighboring sets. When ever data needs to be transmitted from a node in one set to a node in another set, the nodes which connect the sets tend to get overwhelmed, and often result in a bottleneck. This paper proposes a new algorithm for constructing ad-hoc networks in which the distance between nodes and the already constructed directed graph of node connections is used to produce new connections. Nodes are arranged in groups (relatively close to one another) and each group is connected to other groups. The nodes which establish links between neighboring groups alternate with each other as the role of the communicator. This prevents any single node from being overwhelmed with transmissions, and consequently provides a better method of distributing the energy requirements over the entire network. One issue of this seemingly simple idea that was not tackled by this paper is that of complexity and code size. The nodes must know the distance of neighboring nodes and also the structure of the existing network when establishing a connection to the network. The processing of potentially large amounts of information could require unrealistic storage space for resource limited wireless sensor nodes.
Taming the Underlying Challenges of Reliable Multi-hop Routing in Sensor Networks
A. Woo, T. Tony, and D. Culler, Taming the Underlying Challenges of Reliable Multihop Routing in Sensor Networks, ACM Conference on Embedded Networked Sensor Systems (SenSys), November 2003.
This paper analyzes the many different algorithms which govern multi-hop routing in sensor networks. The three aspects of wireless sensor node networking that are addressed in this paper are link estimation, neighborhood table management, and reliable routing protocol techniques. Link estimation is the process of evaluating the integrity/quality of a channel of communication between two nodes. The authors found that a Window Mean with Exponentially Weighted Moving Average (WMEWMA) algorithm was the best balance between simplicity and performance when evaluating the quality of a wireless link. For neighborhood table management, the authors discovered that a “frequency” algorithm give much promise. The “frequency” algorithm works much like the algorithm of a “frequently used cache.” When ever a node communicates with a parent node, the parent node increases a counter variable that corresponds to the child node. If a node attempts to communicate with a parent node and the child node is not found in the parent node’s neighborhood table, then the child is added to a slot in the table in which the counter variable is zero. If no slot is available, then the packet is dropped, and every counter in the table is decremented by one. In regard to reliable routing techniques, this paper claims that “minimum expected transmissions” is an effective metric for cost-based routing. Overall, this paper appropriately and without bias analyzed different algorithms for managing wireless sensor node networks. The findings of this paper do not actually improve upon any existing algorithm, but rather simply inform the reader of which algorithms performed better than others in the tests conducted by the authors.
Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computers
C.E. Perkins and P. Bhagwat, Highly dynamic Destination-Sequenced Distance-Vector routing (DSDV) for mobile computers , SIGCOMM Symposium on Communications Architectures and Protocols (SIGCOMM), September 1994.
This paper addresses the need to provide an algorithm for constructing ad-hoc networks which provide guaranteed loop-free paths to any destination. Destination nodes are nodes which most data is being sent. A destination node stores the topology of the child nodes connected to it. When a child node leaves, joins, or connects to a different parent in the same tree structure, then that child node sends a “new route” message to the destination node. After a small amount of time has passes, referred to as “settling” time, the destination node updates its topology table and broadcasts the new topology to all of its child nodes. Since the wireless network takes the form of a large tree structure, the possibility of creating loops is eliminated. This algorithm may work in theory, but may also be unsuitable for wireless sensor nodes which have power limitations as well as memory size limitations. Each node must know the topology of the network, and when any node is moved, the new topology is communicated to all other nodes, thus causing more network traffic.
Self-configuring localization systems: Design and Experimental Evaluation
Nirupama Bulusu, John Heidemann, Deborah Estrin, and Tommy Tran. Self-configuring localization systems: Design and Experimental Evaluation. ACM Transactions on Embedded Computing Systems, 3 (1 ), pp. 24-60, February, 2004.
This paper focuses on the methods used to construct ad-hoc wireless network for resource limited wireless sensor nodes. The two main methods analyzed in this paper are HEAP and STROBE. HEAP is used in low to medium density networks and attempts to determine optimal locations for placing new nodes in locations with poor localization. This allows the automated formation process of ad-hoc networks to display a certain level of intelligence as the network continually tries to improve itself. STROBE is used in dense networks, and allows nodes to periodically conserve energy by letting a nearby node take over communication. This enables the network to conserve energy by removing redundancy. HEAP was tested in a real world environment which actual hardware. The results were not spectacular but they did show that HEAP could improve the integrity of medium density ad-hoc networks. The STROBE method was simulated in software, thus the results are not as trustworthy as the results of HEAP. It would have been interesting to see the STROBE method implemented in hardware with the results consisting of an analysis on network traffic and/or a comparison of how many transmissions and receptions were performed by each node.
GHT: A Geographic Hash Table for Data-Centric Storage
S. Ratnasamy, B. Karp, S. Shenker, D. Estrin, R. Govindan, L. Yin, and F. Yu. Data-centric storage in sensornets with GHT, a geographic hash table. Mobile Networks and Applications, Vol. 8,No. 4, 2003.
PowerPoint Slides
The nesC Language: A Holistic Approach to Networked Embedded Systems
D. Gay, P. Levis, R. von Behren, M. Welsh, E. Brewer, and D. Culler, The nesC Language: A Holistic Approach to Networked Embedded Systems, ACM Conference on Programming Language Design and Implementation (PLDI), June 2003.
This paper is, for the most part, an explanation of how nesC operates. NesC is a programming language developed for small, resource limited wireless nodes, called motes. A special language is needed for these devices because of the tremendous power and memory limitations imposed upon them. According to this paper, nesC achieves very good power reduction and code size by implementing a module based design. Structuring programs into components allows nesC to include certain components when needed and exclude other components when not needed. This enables nesC to achieve attractive code size reductions, which allows motes to use smaller amounts of memory, enabling them to achieve lower power consumption rates. NesC is also a concurrent programming language, meaning that it allows components to run simultaneously rather than linearly. This type of operation is useful when data needs to be dealt with when it arrives, rather than after another component finishes operation.
EnviroSuite: An Environmentally Immersive Programming Framework for Sensor Networks
L. Luo, T.F. Abdelzaher, T. He, and J.A. Stankovic. EnviroSuite: An Environmentally Immersive Programming Framework for Sensor Networks. ACM Transaction on Embedded Computing System (TECS), 2006.
This paper presents a programming architecture called EnviroSuite. EnviroSuite is a high-level abstraction that allows programmers to think in terms of environmental elements rather than in terms of sensor nodes. EnviroSuite is an object oriented language in which objects represent elements in the external environment. The goal of EnviroSuite is to make sensor node programming easier to develop. While EnviroSuite seems to be able to alleviate some of the low level aspects of the conventional nesC language, end the end, it must be compiled into a nesC equivalent. Certain aspects of the EnviroSuite functionality have already been developed in nesC. The EnviroSuite language offers an interface to the programmer for these precompiled modules. Although this does make programming easier, it also makes it less efficient. Since the nesC code for the modules have already been written, the programmer is not free to make shortcuts or attempt to reduce redundancy. Thus when the EnviroSuite program is compiled into nesC, the code size is not optimized, and thus a trade off between program efficiency and ease of programming must be made.
Mate: a Virtual Machine for Tiny Networked Sensors
P. Levis, N. Patel, S. Shenker, and D. Culler, Mate: a Virtual Machine for Tiny Networked Sensors, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), Dec 2002.
This paper introduces a virtual machine for TinyOS called Mate. Mate is similar to Java in that it stores programs as byte code and then interperets that byte code into machine instructions. The purpose of this VM is so that new programs can be uploaded to each node without causing too much of a power drain on the network. According to this paper, installing new programs on nodes causes nodes to use large amounts of power because the programs being installed are usually quite large. The introduction of a virtual machine that uses byte code would effectively reduce the amount of code needed for programs, thus reducing the amount of energy needed to send the program. This is a good idea, but it has a flaw. The virtual machine will undoubtedly make programs smaller, but it will also introduce overhead when running the programs. The extra cpu cycles will cause the nodes to consume more energy, thus canceling out the benefit of reduced code sizes. The only way that the virtual machine would be beneficial is if the number of program installs was extremely frequent, or the nodes spend a large majority of their life in a form of sleep mode. However, one positive aspect of a virtual machine is that it could allow code to be written once, and place on many different types of hardware, thus programs would be much more portable than they are now.
Active Sensor Networks
P. Levis, D. Gay, and D. Culler, Active Sensor Networks, USENIX/ACM Symposium on Networked Systems Design and Implementation (NSDI), 2005.
This paper presents an improvement upon the Mate virtual machine architecture. This improvement is known as Application Specific Virtual Machine (ASVM). This is an attempt to improve the faults of the Mate VM. The main drawback of Mate was that it required too much overhead. The new concept is to combine the advantages of Mate with the advantages of non VM systems. The advantage of Mate is that it allows code updates to be transmitted across the network in much smaller packets. The advantage of not having a VM is that programs have no overhead when executing. By making application specific virtual machines, that is virtual machines that exist for a specific application, the marriage of the two advantages is possible. The others present several different examples such as a VM database system. If this code needs to be updated often then it benefits from the VM architecture, but if another piece of code does not need to be updated often, then it would benefit from not existing on a VM. VM are thus constructed only for applications which would benefit them. Although this is a nice concept, the authors do not really go into much testing detail. They claim that the energy cost due to overhead is immeasurable because of the low duty cycle of applications. This may be an acceptable assumption, but actual numbers would be nice.
VigilNet: An Integrated Sensor Network System for Energy-Efficient Surveillance
T. He, et al. VigilNet: An Integrated Sensor Network System for Energy-Efficient Surveillance. ACM Transactions on Sensor Networks, 2(1): 1-38, February 2006.
This paper presents a military sensor network used for surveillance in extremely hostile environments which would be too dangerous for human soldiers. The network analyzed in this paper is that of a vehicle tracking system, in which wireless sensor nodes placed alongside a road use sensors to monitor the passing of vehicles. This paper mandates that this network must be effective and “stealthy,” meaning that it should not transmit a lot of information which would in turn allow the enemy to detect the network. The authors mentioned the need for the nodes to have long battery life and thus recommended over population of the nodes in order for the nodes to share surveillance responsibilities, thus spreading the power requirements of a node over the network. One overwhelming drawback of the nodes is there inability to accurately detect fast moving vehicles. According to the authors, vehicles speeds which were in excess of 10 mph introduced erroneous surveillance information into the system. This, of course, must be resolved before the wireless nodes could actually be used in battle.
Design and Deployment of Industrial Sensor Networks: Experiences from a Semiconductor Plant and the North Sea
L. Krishnamurthy, R. Adler, P. Buonadonna, J. Chhabra, M. Flanigan, N. Kushalnagar, L. Nachman, and M. Yarvis, Design and Deployment of Industrial Sensor Networks: Experiences from a Semiconductor Plant and the North Sea, ACM Conference on Embedded Networked Sensor Systems (SenSys), November 2005.
This paper juxtaposed two implementations of wireless sensor networks. One implementation was used on an oil tanker in the North Sea while the other was used in a fabrication house for semiconductors. Both nodes were used to monitor heavy equipment/machinery; however, both systems had different constraints which governed the operation of the wireless network. Two separate wireless node architectures were employed in this study: mica2 and Intel. From the miniscule amount of actual data comparisons between these two platforms, the authors proclaim that the Intel motes provide better power consumption per computation that that of the Mica2. Although the scenarios presented in this paper are interesting, there seems to be a lack of statistical information supporting the mode of the authors in regard to the preference of the Intel motes over the Mica2 motes.
Monitoring Volcanic Eruptions with a Wireless Sensor Network
Geoff Werner-Allen, Jeff Johnson, Mario Ruiz, Jonathan Lees, and Matt Welsh, Monitoring Volcanic Eruptions with a Wireless Sensor Network, EWSN'05.
This paper presents an alternative to cumbersome, power hungry, wired sensors used for volcanic studies. Volcanologists use seismic sensors as well as infrasonic sensors to monitor the activity of a volcano. In order to make data collection easier and less dangerous, the research team developed a wireless sensor node implementation of the current wired sensor network. The sensor nodes were placed at various spots around the volcano and operated in parallel with the existing wired system. The nodes were synchronized using a Garmin GPS receiver which was connected to a distribution node used for long distance communication between the sensor network and the researcher’s base station. Unlike the wired network, the sensor nodes do not store data locally. The data collected by the sensor nodes is transmitted to the base station in real time, thus causing serious network traffic during volcanic events. The nodes were allowed to operate for 54 hours and then the data reported was compared to that of the wired network. Although the information collected by the wireless nodes directly corresponded to the wired network data, some data was lost during transmission. The data rates needed to collect the volcanic information prohibit many nodes from operating at the same time. This limitation led the team to develop a voting algorithm to govern the wireless nodes. Instead of transmitting data continuously, the nodes wait for an event to trigger data collection/transmission. Once an event has been triggered, the nodes communication with their neighboring nodes in order to see if any of the other nodes “heard” the same event. If a certain number of nodes report back with a confirmation of the event, then the node will begin transmitting its data. If not enough nodes report a confirmation of the event, then the event is assumed to be an anomaly, and no data is sent to the base station. This algorithm was intended to relieve network traffic and thus conserve energy, but the results showed that average energy consumption was really about the same.
Model Driven Data Acquisition in Sensor Networks
A. Deshpande, C. Guestrin, S. Madden, J. Hellerstein, and Wei Hong. Model Driven Data Acquisition in Sensor Networks. VLDB 2004.
PowerPoint Slides
The Semantic Web
Tim Berners-Lee, James Hendler and Ora Lassila, "The Semantic Web", Scientific American, May 2001
This paper discusses the future of the World Wide Web. According to this paper, the web will evolve into a medium designed to help humans and machines. The web currently favors humans, as most information on the web is in the form of web pages. Although some web services are geared toward machine computation, the majority of information is not. In order to achieve automation of daily tasks which utilize public information, computers will need a different/universal method for retrieving information. The semantic web is an extension of the already effective world wide web which will be constructed for machine communication. Two methods already exist today which allow machines to retrieve specific information from web services. These two technologies are XML and RFD. Both methods give a specific datum a unique category. As the semantic web evolves, it will be forced to incorporate ontologies which express how different data and categories are related. It is the development of ontologies what will allow machines to interpret data and make human like decisions from information obtained via the semantic web.
The Semantic Web Revisited
Nigel Shadbolt, Tim Berners-Lee and Wendy Hall, "The Semantic Web Revisited", IEEE Intelligent Systems, 21(3) pp. 96-101, May/June 2006
This paper discusses the progress (or lack there of) made since Tim Berners published the paper "The Semantic Web." Although many years have passed since the original paper was published, not much progress has be achieved. There have been various organizations that have started to get the ball rolling, but a disappointing few applications that are actively utilizing the semantic web. The focus of this paper was on ontologies. As in the previous paper, this paper expresses the need for bigger, better, and more abundant ontologies. This paper provides examples of certain scientific communities which utilize ontologies, however makes note that the revolutionary potential of ontologies seems to only be utilized in "scientific" communities. As ontologies begin to be developed in different communities, the cost of developing ontologies is expected to decrease. Actually, the "cost" will increase, but the paper says that the "cost per user" will decrease, since more people will be using the ontologies. This paper is in many ways like its predecessor. It is optimistic towards the end and gives hope for a new semantic web.
OpenGIS Sensor Model Language (SensorML) Implementation Specification
Open Geospatial Consortium, OpenGIS Sensor Model Language (SensorML) Implementation Specification, Document #OGC 05-086r2, Feb. 2006
This paper presents a modeling language for sensors called SensorML. SensorML is used to provide standardized descriptions of sensors, including what sensors do and the process or process-chain needed to utilize sensor functionality. The prized innovation of SensorML is the foundation of its architecture in phenomena detection. Sensors are defined by the types of phenomena they detect. The descriptive language used to characterize the sensors in SensorML is XML. XML, which has taken many data organization and characterization problems by storm is used to describe sensor attributes like type, and quantity. SensorML is a needed application in this rapidly progressing age due to the lack of unification in sensor manufacturers. SensorML could provide a standardized method of sensor descriptions for many future sensors.
Plug-and-play sensors in wireless networks
Dunbar, M., Plug-and-play sensors in wireless networks, IEEE Instrumentation & Measurement Magazine, Vol.4, No.1, pp.19-23, March 2001
This paper promotes the development of wireless standards for peripheral devices. According to this paper, the growth of wireless devices is growing at such a rapid pace, that at some point every serial, parallel, and USB cable will be replaced with a wireless equivalent. The current leader in standardized wireless communication is Bluetooth. The Bluetooth standard ensures that any Bluetooth enabled device will work with any Bluetooth enabled receiver, regardless of manufacture. A current application which needs wireless technology standards is that of sensor networks. In many situations, it is more practical to send information wireless to a data collection device, than it is to run cables. The CrossNet node utilizes the IEEE 1451 wireless standard to collect information from various sensors and send this information wirelessly to a controller node. The use of well defined standards has made this node completely diverse in which it does not discriminate on the type of information being collected. Clearly standards for wireless communication are needed.
Semantics-enabled framework for knowledge discovery from Earth observation data archives
Durbha, S.S.; King, R.L., Semantics-enabled framework for knowledge discovery from Earth observation data archives, IEEE Trans. Geoscience and Remote Sensing, Vol. 43, no.11, pp.2563-2572, Nov. 2005
Review is forthcoming
Minimum Power Configuration in Wireless Sensor Networks
G. Xing, C. Lu, Y. Zhang, Q. Huang, and R. Pless, Minimum Power Configuration in Wireless Sensor Networks, ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), May 2005.
This paper, like many others, brings attention to the never ending problem of power consumption in a wireless sensor network. However, unlike any other paper thus far, this paper uses set theory to find an optimal power configuration of a WSN from the combination of three well known power schemes. These power schemes include topology control, power-aware routing, and sleep management. The algorithms developed in this paper include: Matching based Approximation (MBA), Shortest-path Tree Heuristic (STH), Incremental Shortest-path Tree Heuristic (ISTH), and Constant-ratio Approximation Algorithm. The authors do not go into much detail on MBA and STH, but do chose to implement ISTH in their Minimum Power Configuration Protocol (MPCP). Although an optimized version of MBA actually provides smaller energy consumption per data rate, the ISTH algorithm was used because it was simpler to implement and based off of a shortest path algorithm. MPCP proved to be quite successful and provided 33% less energy consumption over the base protocols of Minimum Transmission (MT) and Minimum Transmission Power (MTP).
Flexible power scheduling for sensor networks. Information Processing in Sensor Networks
Hohlt, B. Doherty, L. Brewer, E., Flexible power scheduling for sensor networks. Information Processing in Sensor Networks, 2004. IPSN 2004. Third International Symposium on, 26-27 April 2004, pp. 205-214
This paper presents a Flexible Power Scheduling (FPS) protocol which utilizes course grain scheduling and fine grain medium access to communication between wireless nodes. The main goal of this protocol is to reduce the amount of energy consumed by a wireless network by allowing nodes to turn off their radios and enter a sleep state. In order to make sure that communication exists between any two groups of nodes, a systematic form of scheduling must be present which tells the nodes when to sleep and how long to remain asleep. In this protocol nodes reserve time slots for communication with other nodes. If a section of time slots is not reserved, then the node is allowed to sleep until the next reserved time slot is reached. A child node sends what the authors call and “advertisement” to a parent node which is a request to reserve a time slot for communication. Since communication only occurs during reserved time slots, time synchronization is of extreme importance.
Low-energy sensor network time synchronization as an emergent property
Bush, S.F., Low-energy sensor network time synchronization as an emergent property. Computer Communications and Networks, 2005. ICCCN 2005. Proceedings. 14th International Conference on. 17-19 Oct. 2005, pp. 93-98
This paper introduces an emergent clock synchronization technique known as Pulse Coupled Oscillation. In this algorithm a master clock node (most likely equipped with a GPS receiver) sends out a frequency of pulses in a 5ms window. This transmission occurs every 500 seconds. The neighbor nodes compare the last pulse of this transmission with their current oscillator pulse. An offset in the two pulses is then obtained to synchronized the receiving node to the time of the transmitting node. The optimal network topology for this algorithm is a single hop from any node to a GPS enabled node. This allows all GPS enabled nodes (which by nature will be synchronized with each other) to synchronized all neighboring nodes in the shortest amount of time possible. The less time it takes to synchronize, the less power is consumed. This paper was slightly vague on the operations of their algorithm and in turn made it difficult to visualize its operation. The simulation performed in this paper assumed each node neighbored a GPS receiver and nodes were randomly distributed with an average transmission distance of 300 meters. It would be interesting to know how this algorithm performs under a topology in which nodes are much closer together and have many more neighbors.
TSync : A Lightweight Bidirectional Time Synchronization Service for Wireless Sensor Networks
H. Dai, R. Han, "“TSync : A Lightweight Bidirectional Time Synchronization Service for Wireless Sensor Networks”, ACM SIGMOBILE Mobile Computing and Communications Review, Special Issue on Wireless PAN & Sensor Networks, vol. 8, no. 1, January 2004, pp. 125-139.
This paper presents a lightweight protocol for synchronizing nodes in a wireless sensor network. TSync offers two methods to achieve synchronization. The first method is referred to as a “push” method used for global synchronization and a “pull” method used for single node synchronization. The push method is performed periodically by a base node (most likely equipped with a GPS receiver for accurate time keeping). The base node sends a global message to all nodes containing its current time value. The receiving nodes can use this value to determine an offset and thus adjust their time values to match as close as possible to that of the base node. The “pull” method is used when a single node wishes to synchronize its clock with that of the base station (or a neighboring node). This method allows a single node to query another node for its current time value and thus upon receiving a reply, adjust its clock to a synchronized value.
Calibration as Parameter Estimation in Sensor Networks
Kamin Whitehouse and David Culler, “Calibration as Parameter Estimation in Sensor Networks”, 2002 ACM International Workshop on Wireless Sensor Networks and Applications, Sept.2002.
This paper introduces a calibration technique cleverly called “calamari.” Calamari is a macro level calibration technique that gathers information from each device in the network and uses the collected data to calibrate each device so that the network is optimized as a whole. This technique showed remarkable improvements over existing techniques such as mean calibration and iterative calibration. This of course would make sense because “calamari” is a macro level technique. Instead of each device setting its own parameters in an attempt to calibrate itself, the controller node is able to see the network as a whole, and thus determine the true optimization for each device. One aspect not mentioned in this paper, however, is that of power. How much power would this technique consume due to the flood of incoming data to the controller node? What about time calibration and latency of transmissions?
