From Ece

Return to the paper reading schedule or the Sensor Networks homepage.

System Architecture Directions for Networked Sensors

powerpoint paper

Connecting the Physical World with Pervasive Networks

In sensor networks, there are many challenges researchers encounter in an effort to revolutionize technological developments. The authors outline many problems faced in connecting the physical world with perspective networks. The first and most important are those that involve systems challenges. The system challenges are the immense amount of distributed system elements, limited physical access to them, and the regime's extreme environmental dynamics. There are drawbacks trying to implement personal digital assistants and pocket PCs in senor networks such as low bandwidth, void of proximity information, very expensive, consumes too much energy, and requires a large battery pack. Several research groups are exploring how to build intelligent, information-rich physical environment by deploying many small, unethered, deeply integrated nodes, however, the challenges involves designing systems to operate at low power consumptions, storing and obtaining that power, orchestrating nodes to form larger networks, and deploying applications over fine-grain networks. Interfacing to the physical world involves exchanging energy between embedded nodes and their environments include the tasks of sensing and actuating, but some common problems are uncertainty and latency. In localization, the problem is to design algorithms that can rationally combine the data from these sources to maintain an estimate of node location.

The authors did propose solutions to these obstacles. For system challenges, the author suggests taking a step towards identifying common building blocks is to define taxonomy of systems and applications so that researchers can identify and foster reusable and parameterized features. In order to overcome the problems associated with personal digital assistants and Pocket PCs, connect rich sensor arrays and sophisticated motor controllers to these devices as would a PC in a traditional process control environment or to an embedded controller. In order to build intelligent, information-rich nodes and to design a system that is operable, the system should be designed to handle asynchronous sensor events, support localized data processing algorithms, and be robust and reprogrammable in the field. Hybrid control should be adapted for solving the problems of sensing and actuating.

The purposes of sensor networks are to have little human interactions as possible. The taxonomy of systems that the authors introduced is the main contribution of this paper. Taxonomy includes the areas such as scalability, variability, and autonomy. Localization, which is a major area of research in sensor networks, sensing, and actuation are also technical strengths that the authors introduces in this paper because nodes have to interact with each other as well as the physical environments in which they are deployed but with little or no human intervention. However, the authors fail to propose a solution for an energy efficient sensor network. Energy consumption is an ongoing problem in systems that are deployed because without an energy efficient method nodes functionality will be unsuccessful and the expense to deploy more nodes will increase due to its high demand especially in hazardous environments.

In the area of energy consumption, the paper could have included a discussion on the importance of sleeping and idle nodes. The paper could have included the issue of communication between nodes that are deployed in physical environments. The topics of ad hoc and multihop should have been introduced by the authors in this paper as well. However, for this paper to have been written in 2002 when there wasn’t substantial research being performed by many researchers, the authors did provide a very informative paper.

Sensor Networks for Emergency Response: Challenges and Opportunities

The major problem that researchers are experiencing with sensor networks is that sensor nodes have extreme resource limitations in protocol design, application development, and security models. Senor nodes are extremely limited in communication and computational capabilities. Amongst the obvious concerns with sensor networks, there are other issues that arise in these systems. The main issues are discovery and naming, robust routing, prioritization of critical data, security, and tracking device locations. Researchers have proposed methods to help solve these key issues. For discovery and naming, researches believe that is if the discovery process is decentralized then this will help avoid any single point of failure. To enhance the robustness of routing protocols, ad hoc routing techniques should be implemented to extend an effective communication range by having devices relay messages for one another. When many devices share this bandwidth, the critical data should be given priority over other traffic, and by employing this method the issue of prioritization of critical data is solved. Using many wireless devices such as GPS, RF signals, and ultrasounds will help with location awareness. Security is an essential concern with many sensor networks so having an efficient establishment of security credentials will significantly help secure such systems.

The CodeBlue infrastructure integrates sensor nodes and other wireless devices into a disaster response setting and provides facilities for ad hoc network formation, resources naming and discovery, security, and in-network aggregation of sensor-produced data. CodeBlue is designed for rapidly changing, critical care environments. Nevertheless, MoteTrack is a RF-based location tracking system specifically designed for disaster response. It operates using the low-power, single chip radio transceivers founding networks nodes, which rescue personnel can easily wear or embed in wearable vital sign sensors. These two methods are strongly emphasized due to the fact that using these kinds of protocol and software in disaster response scenarios will help track and monitor patients and responders.

The issue of a lost of batter power during rescue missions needs to be further examined. How much are these devices going to cost the medical community to employ in their field?

The authors should strongly emphasize more about the security attacks and countermeasures associated with the CodeBlue and MoteTrack systems. Security is a key issue in any sensor networks where outside influence can be embedded. Hostile threats can come from anywhere at anytime and security features have to be able to protect the system from any breaches that may occur. Privacy is what each job is to provide to the end user.

Habitat Monitoring with Sensor Networks

Using sensors to monitor various habitats, there are many factors that can affect the research being performed. Sensors must be unobtrusive yet durable under a range of environmental stresses, including damage caused by the organisms themselves. They must be so energy efficient that they can remain in situ wit little human interactions and be maintenance-free for years at a time. Sensors must also be reliably interconnect with a cyber infrastructure that permits frequent network access for data upload, device programming, and management. In addition, organisms can also alter their surroundings in important ways. For example, tree shape, physiology, and canopy structure can produce significantly different ranges of light, humidity, and temperature than might occur in an adjacent open area. Also, burrow-nesting birds, insects, or mammals might create a further unique range of climatic values through their nest-chamber construction. Researchers could get a better understanding of population dynamics, morality factors, and habitat need if scientists scale their data collection to match the full range of an organism’s activities. In addition, pushing computation into individual devices can reduce the energy consumption and volume of data being communicated which will help to improve energy efficiency.

The main contribution the authors emphasize is the system architecture for habitat monitoring. The routing service in habitat monitoring networks delivers the queries to the sensors nodes and reports the data of interest; that data is either streaming or triggered. The percentage of time each node is awake is known as the node’s duty cycle. Lastly, the architecture is made up of a component called network health monitoring. A health monitoring service is critical to providing performance and status information to remote administrators over the life spans of these systems, allowing users to perform maintenance and estimate the confidence associated with the readings.

In order to enhance habitat monitoring networks there should be advances in robust localization, calibration, clock synchronization, and data processing. This technical paper should have includes such advancements in these areas.

The authors should have discussed the affects of data gathering. What does this information obtained through habitat monitoring tell researchers about issues affecting animals, plants, and people?

Towards a Sensor Network Architecture: Lowering the Waistline

The challenge in sensor networks is that their modes of operation introduce requirements and trade offs that are very different from traditional systems. In order to define sensor network architecture, it is difficult to do this because researchers have to decide what to specify in its services and what to leave open. In addition, complete sensornet architecture will need to address a family of specific issues, such as discovery, topology management, naming, routing and so on, but the over-riding question is whether there is “narrow waist”- a unifying abstraction that permits a wide variety of uses above and a range of implementations below.

SP (Sensor-net Protocol) is the keystone of the authors sensor network architecture which will bridge higher level protocols and applications to underlying data link and physical layers.

The technical weakness that the authors have in this paper is the concept that with the given adversaries there aren’t any proposed solutions to help overcome such challenges. Proposing or surveying valid solutions for the challenges of developing this architecture would have made this architecture more appealing to researchers.

Medium Access Control with Coordinated Adaptive Sleeping for Wireless Sensor Networks

Energy consumption is a major issue in wireless sensor networks. Researchers are having a difficult time developing protocols that will prolong the network lifetimes for nodes. Even the current MAC protocols aren’t energy efficient, they too fail to prolong the lifetime of a sensor networks. The development of Senor-MAC (S-MAC), which is a MAC protocol, has researchers on the end of their seats because S-MAC proposes to reduce energy consumption while providing good scalability and collision avoidance.

The challenges that S-MAC addresses are reducing energy waste, idle listening, collision, control overhead, overhearing, and latency.

In order to reduce idle listening, S-MAC implements a technique called low-duty cycle scheme in multihop networks. It reduces idle listening by periodically putting nodes into a sleep state. Instead of node being idle and wasting energy they sleep for some time and the wakes up and listen to see if another node wants to communicate.

To reduce control overhead and latency, S-MAC introduces coordinated sleeping among neighboring nodes. The nodes choose and maintain a schedule and synchronization.

For collision, S-MAC follows similar procedures as 802.11 which include virtual and physical carrier sense and the RTS/CTS exchange for hidden terminal.

S-MAC tries to eliminate overhearing by letting interfering nodes go to sleep after they hear and RTS or CTS packet.

The technical strengths of this paper are the ideas that were proposed to solve the problems of energy efficiency were the experimented and implemented on the Rene Motes and Mica Motes. Also the experiments used to measure and compare energy, latency, and throughput. However, there are tradeoffs on energy, latency, and throughputs.

The S-MAC authors were very informative and presented well organized and execute data that make using S-MAC protocol more feasible than using the 802.11 protocol. As long as its user friendly and open source S-MAC will probably dominate any other protocol.

MACAW: A Media Access Protocol for Wireless LAN’s

As technology increases with wireless devices, indoor wireless local area networks are expected to improve with the curve. The authors of this paper propose a method called MACAW, which uses and RTS-CTS-DS-DATA-ACK message exchange, which is a spin-off of the MACA.

The challenges that the paper mention is contention is at the receiver; not the sender, congestion is local dependent; to allocate media access fairly, learning about congestion must be a collective enterprise; media access protocols should propagate synchronization information about contention periods.

The backoff algorithm used by the MACA protocol had been modified by the MACAW. The MACAW’s backoff algorithm includes in the packet header a field which contains the current value of the backoff counter which makes the throughput allocations fair. Also the RTS-CTS-DS-DATA has been modified from the original. The MACAW’s RTS-CTS-DS-DATA includes an acknowledgement packet, ACK, which is returned from the receiver to the sender immediately upon completion of data reception. The two new methods are helping to create short contention periods and ease the congestion in LAN networks.

Many issues were addressed to be solved, but the authors have not discovered any solutions to these ongoing problems. There will probably be more research being performed on the MACAW protocol in the future because this protocol has much future work that needs to be researched.

Ultra-Low Duty Cycle MAC with Scheduled Channel Polling

Similar to the S-MAC, energy consumption is a major issue in wireless sensor networks. This paper proposes a new method that piggybacks off S-MAC called scheduled channel polling (SCP) which should help achieve ultra-low duty cycles.

SCP uses to approaches to help solve the problem of energy consumption. One approach is using scheduled protocols. A second approach is performing low-power listening. Implementing these two methods collectively will help to reduce energy consumption and create a system that operates on an ultra-low duty cycle. However, the key to achieving this type of protocol is synchronization of all the nodes. The nodes scheduling, transmissions, and listening all have to be synchronized to maximize the use of energy.

This paper piggybacks off different protocols already being utilized in sensor networks today. Using the methods to model yet enhance performance was a sufficient method at trying to solve the issue of energy consumption. This method in itself has a lot of future research entailed with the development, implementation, and execution of this protocol on different platforms.

On Calculating Power-Aware Connected Dominating Sets for Efficient Routing in Ad Hoc Wireless Networks

The original algorithm used by Wu and Li [1] proposed a simple and efficient distributed algorithm for calculating connected dominating set in ad hoc wireless networks, where connection of nodes are determined by geographical distances of nodes. This paper proposes a method that piggybacks off Wu and Li algorithm but with a different goal at hand. Instead of calculating power-aware connected dominating set based on the geographical distance of nodes, this algorithm calculates power-aware connected dominating set based on a dynamic selection process.

This algorithm being implemented is focused on solving the problems of prolonging the life span of each node and the networks, and also, balancing the energy consumption in the network. The algorithm uses two extensions to help with these problems. One is based on node degree which aims at reducing the size of the connected dominating set generated from the marking process. On the other hand, the second approach based on energy level focuses on prolonging the average life span of a host, and at the same time, reducing the size of a connected dominating set generated from the marking process.

These authors measure the size of the connected dominating set generated from the marking process and compare it with the size of the connected dominating set after applying different rules, which include the rules based on ID, the rules based on ND, and the rules based on EL. In addition, the average life spans of the network under different rules are also simulated. The results have shown that the proposed approach based on energy level is clearly the best in terms of the longer life span of the network.

The paper mentions the contentions and collision, but fails to address the importance of how these two aspects affect energy consumptions. One of the goals of this paper is to balance the energy consumption so it would have been important to address prolongs life span of a host when challenged with either a collision or contention.

Taming the Underlying Challenges of Reliable Multihop Routing in Sensor Networks

The authors of this paper focus on finding an estimator that reacts quickly to potential large changes in link quality, yet is stable, has a small memory footprint, and is simple to compute. Another concern of this paper is developing a neighborhood table manager that inserts, evicts, and reinforces links. Lastly, designing a global network infrastructure is a challenge that the authors address.

A protocol called WMEWMA was introduced because it is a simple, memory efficient link estimator that reacts quickly yet is stable enough for path characterization in cost-based routing. The FREQUENCY algorithm performs well in maintaining a subset of good neighbors in a constrained neighbor table regardless of cell density.

The underlying problems that the authors address and propose a solution for are still a concern for researchers. This paper did many evaluations, but failed to help close the gap on link estimation, neighborhood table management, and developing a global infrastructure.

Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computers

This paper introduces the need for designing an ad-hoc networks and it develops a protocol which help utilization of the ad-hoc network more efficiently.

Designing an Ad-Hoc network is the primary motivation for this paper. An ad-hoc network is the cooperative engagement of a collection of Mobile hosts without the required intervention of any centralized access point. Destination-Sequenced Distance-Vector Routing (DSDV) protocol was developed to assist in developing routing in an ad-hoc network that provides the kind of dynamic, self-starting behavior needed in such a network. DSDV routing protocol allows a collection of mobile computers, which may not be close to any base station, to afford all computers among their number a path along which data can be exchanged. DSSV protocol requires each mobile station to advertise to each of its current neighbors and its own routing table. Also this protocol is used to extend base station coverage beyond range of transmission. DSDV can be utilized at either the network layer or below the network layer but still above the MAC layer.

This issues that it didn’t address that would have been relevant is the issues of collisions or redundancy. What does the DSDV routing protocol do when a collision occur or redundancy happens in the network? What if jamming happens in the networks does DSDV solve that problem?

Self-Configuring Localization Systems: Design and Experimental Evaluation

The focus of this paper is to address the issue of creating a localized sensor network that is self-configuring. In order to achieve this kind of systems, the authors propose that beacons and beacon density are key approaches to the success of an adaptive, localized sensor network. There are two algorithms introduced for system self-configuration.

The first algorithm introduced was the HEAP algorithm to detect regions with poor localization, and select candidate points for placing new beacons. However, a key requirement for large-scale sensor networks is robust, unattended operation. Therefore, for dense beacon deployments, the authors propose the STROBE algorithm. STROBE enables densely deployed beacons to coordinate without self-interference and opportunistically conserve energy.

The authors evaluated both algorithms but in two different environments. The HEAP algorithm was evaluated using Berkeley Rene mote in a real environment to verify it effectiveness. On the other hand, to evaluate STROBE the authors chose an energy consumption model to mimic realistic sensor radios. The HEAP was realistically used and evaluated which produces better results because of the real-time applications and computations. Yet, the STROBE wasn’t evaluated in real-time, and the authors approximated energy consumption. This is an issue because when motes are deployed the energy consumption is a key factor in the life span of a network. The STROBE algorithm should be reevaluated on real time applications, and the energy consumption, being important as it is, should also be computed and not assumed.

A Unifying Link Abstraction for Wireless Sensor Networks

http://www.ece.msstate.edu/~ssl15/networks/PaperPresentationFeb19.ppt powerpoint 

The nesC Language: A Holistic Approach to Network Embedded Systems

nesC is a programming language for networked embedded systems that represents a new design space for applications developers. nesC is also an extension of C language. nesC supports programming models that integrates reactivity to the environment, concurrency, and communication.

nesC provides three brad contributions. First, nesC defines a component model that supports event-driven systems. Second, nesC defines simple but expressive concurrency model coupled with extensive compile-time analysis. Third, nesC provides a unique balance between accurate program analysis to improve reliability and reduce code size, and expressive power for building real applications.

nesC is a static language which make whole programs analysis and optimization significantly simpler and more accurate.

For embedded networks, nesC is a good idea as of now. However, will nesC be dominant when embedded networks dramatically improve over the years with the given technologies? Is being static such a pro or is it a con for nesC? What is the highest or lowest level nesC can be implemented?

EnviroSuite: An Environmentally Immersive Programming Framework for Sensor networks

Implementing paradigms for sensor network programming in the large have been identified as a significant challenge towards developing large-scale applications. The classical programming languages are too low-level to develop large-scale applications.

EnviroSuite is a programming framework that introduces a new paradigm, called environmentally immersive programming, to abstract distributed interactions with the environments. EnviroSuite is not another programming language introduce to replace other languages. EnviroSuite is a framework that extends other programming languages with environmental immersive programming primitives which consists of a list of object declarations. EnviroSuite is compiled through a compiler called EIPLC and then translated into nesC language.

If nesC is a static language and in EnviroSuite objects are instantiated either statically or dynamically then is nesC instantiated statically or dynamically? Or does the EIPLC compiler negotiate between the two?

Mate’: A Tiny Virtual machine for Sensor Networks

Sensor nodes are deployed in various environments each day. Accessibility to these nodes is almost infeasible. Therefore, sensor nodes must be reprogrammable in response to changing environments. This paper introduces mate’. Mate’ is described as a tiny communication-centric virtual machine designed for sensor networks.

Mate’ is a byte-code interpreter to run on TinyOS. Code is broken up into capsules of 24 instructions. These capsules can forward themselves through a network with a single instruction. This main contribution of Mate’ Virtual machine is the idea of code being broken up into capsules which can self-replicate through the network.

Ad-hoc networking is critical system issues in sensor networks, this paper introduce an algorithm called BLESS to better help address this issue. BLESS includes routing information in every packet and transmits everything on an AM broadcast. Implementing the BLESS algorithm, the authors demonstrate that Mate’ can transform sensor networks into active networks.

This paper didn’t propose a rough figure of how much this will cost to implement Mate’ to run on TinyOS. Will Mate’ be user friendly when writing code or will be complicated like writing code in TinyOS?

Active Sensor Networks

Active Sensor Networks is a spin-off of the Mate’ virtual machine but taken to a higher level. This paper proposes using application specific virtual machines (ASVMs) to reprogram deployed wireless sensor networks. Rather than propose a new programming approach to in-network processing, ASVM is a propose architecture for implementing a programming model’s underlying runtime.

ASVMs have three major abstractions: handlers, operations, and capsules. Handlers are code routine that runs in response to system events, operations are the units of execution functionality, and capsules are the units of code propagation. Just like TinyOS, ASVM is a simple FIFO thread scheduler.

An evaluation of ASVM showed that an ASVM can issue just under ten thousand instructions per second on a 4Mhz mica. The ASVM decomposition imposes a 6% overhead over a similar loop in Mate’.

The authors have improved since the design of Mate’ because Mate’ has predefined set of three events it executes in response to there are RAM constraints limiting code for a particular event handler to 24 bytes long.

A Wireless Sensor Network for Structural Monitoring

http://www.ece.msstate.edu/~ssl15/networks/PaperPresMar5.ppt powerpoint

Chord: A scalable Peer-to-peer Lookup Protocol for Internet Applications

Chord provides support for just one operation: given a key, it maps the key onto a node. Chord uses consistent hashing to assign keys to Chord nodes. Consistent hashing tends to balance load, since each node receives the same number of keys, and requires relatively little movement of keys when nodes join and leave the system. This look-up protocol simplifies the design of peer-to-peer systems and applications by addressing the problems as load balance, decentralization, scalability, availability, and flexible naming. Chord nodes require information about 0(log N) other nodes for efficient routing, but when information is out of date the performance degrades gracefully. When a node joins the network, certain keys that were assigned to its successor are now assigned to the new node. Also, when a node leaves any keys that were assigned to that node are now assigned to its successor.

The Sensor Network as a Database

The individual nodes in sensor networks do not necessarily have an identity of interest; rather, the data that they generate is of interest independents of source node identity. The sensornet database allows any user to issue a query to the sensor network as if it is a database system and obtain a response to that query. The architecture rests on two features. The first is in-network implementation of database operators. When a user poses a query to the network, that query is disseminated across the network. The second feature is approximate results. The goal of the sensornet database design should be to preserve location transparency.

Beyond Average: Toward Sophisticated Sensing with Queries

This paper introduces a designed framework called TAG for sensornet database via an SQL-like language. TAG framework is implemented on a system called TinyDB that runs in networks of TinyOS-based Berkeley motes. As the query is flooded through the network, sensors organize into a routing tree that allows the basestation to collect query results. The flooding method operates as following: the basestation injects a query request at the root sensor which broadcast the query on its radio; all child nodes that hear the query process it and rebroadcast it on to their children, and so on, until the entire network has heard the query.

Synopsis Diffusion for Robust Aggregation in Sensor Networks

Synopsis Diffusion (SD) is a general framework for achieving significantly more accurate and reliable answers by combining energy-efficient multi-path routing schemes with techniques that avoid double-counting. SD avoids double-counting through the use of order-and duplicate-insensitive (ODI) synopses that compactly summarize intermediate results during the in-network aggregation. SD consists of two phases: a distribution phase and an aggregation phase. Using synopses may provide only an approximate answer to certain queries where in the final answer there are two distinct errors in the final answers. The first one is the communication error, and the second one is the approximation error.

SD reduces answer errors in lossy environments. Also, synopsis diffusion helps address the challenges imposed by correlated node failures. Finally, SD can achieve these gains without a significant increase in power consumption.