From eyh5@ee.cornell.edu Mon Nov 12 15:46:45 2001 Return-Path: Received: from memphis.ece.cornell.edu (memphis.ece.cornell.edu [128.84.81.8]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fACKkiR23545 for ; Mon, 12 Nov 2001 15:46:44 -0500 (EST) Received: from photon.ece.cornell.edu (photon.ece.cornell.edu [128.84.81.138]) by memphis.ece.cornell.edu (8.11.6/8.11.2) with ESMTP id fACKjNH03022 for ; Mon, 12 Nov 2001 15:45:23 -0500 Date: Mon, 12 Nov 2001 15:45:33 -0500 (EST) From: Edward Hua X-X-Sender: To: Subject: 615 Paper # 42 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII TinyOS: An Operating Systems for Networked Sensors Jason Hill, et al. The group of researchers at UC Berkeley have proposed the TinyOS, an event-driven operating environment designed for use with embeddednetworked sensors. TinyOS is designed to perform with ultra-low power consumption and to support concurrency-intensive applications, i.e., applications who accept only the most recently gathered data and discard the others. TinyOS serves as a system-level bridge between the low-power RF technology and MEMS (Micro-Electro-Mechanical-System) technology. Its attributes include is small physical size, its support for concurrency-critical applications, its diversity in design and usage, and its great autonomy in establishing the sensor network without oversight. In addition, instead of developing the TinyOS from scratch, all hardware components of this platform are purchased off-the-shelf. This makes it very cost-effective. TinyOS takes advantage of the ad hoc sensing technique commonly associated with MANETs. All nodes in a TinyOS sensor network are autonomous, and, once deployed, communicate with each other to assemble the network. One of the nodes is designated as the central collection point, or the base station. The data captured by the nodes of the network propagate to the base station. The information on the connectivity and error rates exchanged between the nodes may be used to infer the relative distance amongst them. A TinyOS, in a sense, is a finite state machine (FSM). A single node functioningly independently is considered a state, and a network of nodes working together represents transitions from state to state. Such a design philosophy allows quick, low-overhead, non-blocking state transmissions, and allows independent components of the network to share a single execution context. This enhances the network's ability to scale flexibly. Another feature of TinyOS is that it blurs the boundary between HW and SW. In fact, the component model used by TinyOS translates and propogates the HW abstractions into software. On the otherhand, it also allows for migrations of SW components into HW. An example is that the bit level radio procession component could be implemented as specialized FIFO with complex pattern matching techniques. This feature makes TinyOS to be highly modular, as it permits a node in the network to transport its software modules to other nodes with minimum effor, therefore achieving application-specific functionality. The performance results of TinyOS are demonstrated in some of the presentation slides made by the Berkeley team. One of the important contributions is that it extends the battery life span, often a critical constraint in sensor network deployment, to nearly a year if running low-load tasks, because it uses the commercially available Panasonic CR2354 560 mAh lithium battery. However, this economic battery consumption is subject to the periodicity of the sensor node listening modes. When the node listens full time, the battery life is about 72 hours. Therefore, the type of application and its requirement of active listening node often determines the life span of the sensor node network. Indeed, the researchers argue that in deploying the sensor network, the paramount concern is the energy consumption, and it is to the end of reducing as much energy consumption as possible that TinyOS is devised. ------------------------------------------------ Smart Dust: Autonomous Sensing and Communication in a Cubic Millimeter Kris Pister, John Kahn, Bernhard Boser This team of researchers at Berkeley are working on the "smart dust," a complete sensor/communication system can be integrated into a cubic millimeter package. Smart dust may find applications in areas such as weather/seismological monitoring, chemical/biological sensors, product quality monitoring, and smart office space. The current technology allows researchers to achievetens of cubic millimeters sensors, called motes, although a breakthrough was expected last July that would push the physical size down to around one cubic millimeter. The source of the energy that supports the functioning smart dust is either solar energy or combustion/thermopiles. The researchers claim that their smart dust may last up to two years when running at 1% duty cyle. The idea behind the smart dust is as follows. Because of the miniscule size of each mote, conservation of power consumption is the top priority. Therefore, a node does not remain on at all times. At any given time, only a few timers and a clock are running while most parts of the mote are in sleep mode. Each timer has a specific, corresponding function to control. When one timer expires, a function module is awakened and begins execute its pre-assigned tasks. The function module remains on for a certain period of time, at the end of which it returns to sleep with its timer reset to begin counting again. A mote is also usually equipped with a receiver, which allows it to communicate with the neighbors of the mote in case of packet transmissions. It too functions in the aforementioned manner in order to conserve energy. The receiver has two major functions. First, it may receive instructional packets from another mote that instruct it to perform some certain task. Second, it may act as a station to help relay an incoming packet or packets to the intended destination. Another energy-conservation technique used in smart dust is through the use of corner-cube-retroreflector. The corner cube retroreflector transmits information just by moving a mirror and thus changing the reflection of a laser beam from the base station. This technique is substantially more energy efficient than actually generating some radiation, as the only energy consumed is that to tweak the orientation of the mirror. One implicit assumption in the research of smart dust is that the base station where all the data collected from the motes goes to must be in the vicinity of the network. This is so because the transmission range of the motes is maintained at a minimum, barely enough for motes to communicate with each other. There are scenarios in which the motes may be spread over in a remote area. In that case, either a data retrieval device with more powerful signal reception power needs to be installed permanently or some kind of fly-over of a airplane-carried receiver is required. In either case, there is a closely bonded proximity that exists between the receiver of the data and the smart dust network. This may be a constraint in finding a suitable application for the smart dust technology. From viran@csl.cornell.edu Mon Nov 12 23:33:53 2001 Return-Path: Received: from moore.csl.cornell.edu (moore.csl.cornell.edu [132.236.71.83]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fAD4XpR17441 for ; Mon, 12 Nov 2001 23:33:51 -0500 (EST) Received: from localhost (viran@localhost) by moore.csl.cornell.edu (8.11.3/8.9.2) with ESMTP id fAD4XkT51057 for ; Mon, 12 Nov 2001 23:33:46 -0500 (EST) (envelope-from viran@moore.csl.cornell.edu) X-Authentication-Warning: moore.csl.cornell.edu: viran owned process doing -bs Date: Mon, 12 Nov 2001 23:33:46 -0500 (EST) From: "Virantha N. Ekanayake" To: Subject: 615 Paper 42 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII TinyOS is lightweigt multithreaded (microthreaded) operating system for wireless sensor environments. They provide a high degree of concurrency to keep the data flowing and reduce the requirements for data buffers. The multi-threading has a two-level scheduling scheme: one low-level for scheduling hardware events, and another for long-running tasks that can be interrupted. The system is event-based and thus does not have the overhead of context switches in a stack based threaded system. The main points of the OS can be described as follows: Components are hierarchical and range from low-level hardware modules to high-level application modules. Commands are used to communicate with lower-levels in a non-blocking manner. Events are generated by hardware components and other lower levels and propagate upwards. Tasks are the main method of performing work and encapsulate command and event generation, and in this system, they run to completion. The task-scheduler is a simple FIFO based system. The OS is very well suited for a micro-controller environment -- it fits in 172 bytes of memory and context switches in 12.75us (on a 4Mhz processor). Active Messages are the most significant introduction to TinyOS -- the example they present shows how they can be used to set up multi-hop routing in an efficient manner. The active message concept is basically a remote procedure call (each message invokes a local handler in the receiver) without all the overhead of a system call. In the TinyOS case, they can be handled via the normal event handler. All in all, TinyOS presented an interesting programming viewpoint towards the design of lowcost sensor networks, something absent from other more hardware centric designs that never really addressed the feasability of using the hardware. Smart Dust describes millimeter scale autonomous wireless sensor devices. They base the sensor concept on MEMS and connect to a custom built processor and communication device. They did build a 100 square millimeter device that was non-functioning back in 1999 -- the current state of affairs isn't evident on their web page. However, they have been leveraging the TinyOS on larger scale devices with RF, laser, IR communication methods. It looks like a promising effort in sensor networks, and they seem to have a unified SW/HW approach that's necessary for a successful implementation. From c.tavoularis@utoronto.ca Tue Nov 13 02:42:20 2001 Return-Path: Received: from bureau6.utcc.utoronto.ca (bureau6.utcc.utoronto.ca [128.100.132.16]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fAD7gIR07843 for ; Tue, 13 Nov 2001 02:42:18 -0500 (EST) Received: from webmail4.ns.utoronto.ca ([128.100.132.34] EHLO webmail4.ns.utoronto.ca ident: IDENT-NOT-QUERIED [port 46605]) by bureau6.utcc.utoronto.ca with ESMTP id <239158-28237>; Tue, 13 Nov 2001 02:42:10 -0500 Received: by webmail4.ns.utoronto.ca id <164259-209>; Tue, 13 Nov 2001 02:42:02 -0500 To: egs@CS.Cornell.EDU Subject: 615 PAPER 42 Message-ID: <1005637308.3bf0cebc0c3d4@webmail.utoronto.ca> Date: Tue, 13 Nov 2001 02:41:48 -0500 (EST) From: c.tavoularis@utoronto.ca MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit User-Agent: IMP/PHP IMAP webmail program 2.2.3 The most recent goal of sensor networks, as seen in the implementation of SmartDust and TinoOS, is to make tiny yet effective ‘motes’ that can handle computing and communication. Sensor networks have many useful applications. They serve the military for surveillance and monitoring. They can monitor environmental conditions such as temperature and humidity from which statistics can be gathered for things such as animal or plant behavior or product quality. Other functionality includes automatic inventory and virtual keyboards. The best way for sensor network technology to accommodate these tasks is to have discrete nature, low cost and easy deployment. SmartDust and TinyOS attempt to reach the extremities of size reduction, as small as a square-millimeter, using inexpensive off-the-shelf hardware components. Sensor networks achieve easy deployment by nature of self- initializing multi-hop routing. A key to the functionality of SmartDust is preserving energy. Preserving energy prolongs the life of individual sensors and the network as a whole, and reduces cost. This is particularly aided by the micro-controller that also controls power consumption. In fact, much of the mote is powered off most of the time, while a few timers allow parts of the mote to power on to perform a task and power off again. There are timers to control only the major parts of the mote including the micro-controller and the receiver. Motes happen to use lasers for communication in a particularly energy efficient manner. Motes simply reflect a laser generated by the base station in an appropriate direction to transmit data to any other mote in the network. The motes are also adaptable and can accept code via the wireless communication interface that will change their functionality. TinyOS is an event-based environment that handles concurrent operations at each of the nodes in a sensor network. TinyOS must also minimize overhead, and does so by managing non-blocking states for each module with fine-grained concurrency. It similarly uses inexpensive off-the-shelf hardware. Both the aforementioned systems achieve impressive results. It is interesting to compare and contrast the two implementations, since one is software oriented and the other hardware. From ranveer@CS.Cornell.EDU Tue Nov 13 10:54:41 2001 Return-Path: Received: from exchange.cs.cornell.edu (exchange.cs.cornell.edu [128.84.97.8]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADFsdR06497 for ; Tue, 13 Nov 2001 10:54:39 -0500 (EST) content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Subject: 615 PAPER 42 X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0 Date: Tue, 13 Nov 2001 10:54:39 -0500 Message-ID: <706871B20764CD449DB0E8E3D81C4D430232E6B0@opus.cs.cornell.edu> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: 615 PAPER 42 Thread-Index: AcFsW4AsjMAIGgsBRDeggJmu7stf/Q== From: "Ranveer Chandra" To: "Emin Gun Sirer" Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from base64 to 8bit by sundial.cs.cornell.edu id fADFsdR06497 TinyOS and Sensor Dust The small size of sensor devices have made the deployment of sensor networks feasible. However, these networks are characterized by nodes with limited battery power, bandwodth, computation and storage. Traditional communication hardware and protocols are not constrained by power and bandwidth, nor by communication and storage. Therefore an efficient sensor network would require a completely new architecture with a rebuild on the communication stack. Research at UC Berkeley has focussed on these two problems: TinyOS attacks the problem in software trying to optimize code and protocols to be power and bandwidth efficient, and smart dust approaches the same problems in hardware. Smart Dust proposes a new look at the hardware for sensor networks, including an optical communication scheme. Reflectors are used to reduce the energy concuption. However, an optical communication would require line of sight reachability among sensor nodes. Overall, the approaches proposed are fascinating. They highlight the fact that traditional RF communication schemes are expensive and alternate schemes should be explored. TinyOS looks at the communication stack for sensor devices, and tries to rebuild the software and protocols for efficiency. A new MAC protocol is proposed for sensor devices. This MAC is specialized CSMA when the traffic pattern is correlated. An adaptive rate control scheme is proposed to reduce the number of collisions. Further, the RTS/CTS/DATA/ACK scheme is modified, so that ACKs are not explicitly needed. Since the data will be forwarded, ACKs are assumed to be free by listening in promiscuous mode. Another novel idea in TinyOS is the use of Active Messages to reduce the energy consumption. This could be used for overlap of computation and communication. The event-driven nature of TinyOS could be used to conserve energy. From ramasv@CS.Cornell.EDU Tue Nov 13 11:53:04 2001 Return-Path: Received: from exchange.cs.cornell.edu (exchange.cs.cornell.edu [128.84.97.8]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADGr2R14835 for ; Tue, 13 Nov 2001 11:53:02 -0500 (EST) content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Subject: cs615 PAPER 42 X-MimeOLE: Produced By Microsoft Exchange V6.0.4712.0 Date: Tue, 13 Nov 2001 11:53:01 -0500 Message-ID: <706871B20764CD449DB0E8E3D81C4D4301E7F28F@opus.cs.cornell.edu> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: cs615 PAPER 42 Thread-Index: AcFsY6gFiBCyA+r7TVWUA8FqMs7GjA== From: "Venu Ramasubramanian" To: "Emin Gun Sirer" Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by sundial.cs.cornell.edu id fADGr2R14835 Tiny Networked Sensors: Motes, Sensor Dust, and TinyOS Small networked devices with sensing capability are going to widely deployed and used in future. This project highlights one of the most important characteristics of such devices - their variability. Sensors could come in all shapes and sizes: for example, motes (few cubic millimeters), Ninjas (few cubic inches) etc. Sensors also have wide variation of capacities in terms of networking harware, processor limitations, power consumption. There are further constaints from the kind of enviroments they are deployed in for example: forest fire, sea bed, building, etc. that affects the mobility, bandwidth availability and power availability. The TinyOS project concerns itself mainly with handling sensors with RF radios and 8-bit processor deployed in a static setting. The presence of a processor and other devices demands an operating system and the embedded nature of the CPU necessitates an OS that is effecient in terms of space, time and power consumption. TinyOS is an operating system that conforms to these requirements. TinyOS presents two classes of execution - tasks and events. Tasks are like processes that have state, stack and computation associated with them and deployed by a scheduler. Events are asynchronously generated signals that can pre-empt tasks. Device interfaces as well as software components modeled to have state, tasks and signals associated with them. This kind of component based interface allows the OS to be ported to different kinds of hardware by changing just a few components. One of the drawbacks of this system is the compile time binding of interfaces that prevents on-line addition of applications and necessitates compilation and reloading upon changes in components. In the absence of a wall outlets most of the sensor devices would be expected to depend on battery power. Thus power forms a significant constraint and communication hardware often tends to consume a most of the power. It is interesting to see to interesting designs to tackle this problem and prevent excessive power consumption. Motes reflect laser light creating light flashes that can be interpreted as bits. While the throughput of this scheme may not be very high, it does end up using practicaly little power to communicate. However line-of sight limitations and the need for a nearby laser source might limit the use of this technique. Bit level modulation can however also end up conserving a lot of power as the transmission power depends on packet size. Using 1 bit messages for periodic beacons allows this network to funciton for long periods of time. Such modifications to hardware might prove quite beneficial in the long run. From teifel@csl.cornell.edu Tue Nov 13 12:02:57 2001 Return-Path: Received: from disney.csl.cornell.edu (disney.csl.cornell.edu [132.236.71.87]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADH2tR16247 for ; Tue, 13 Nov 2001 12:02:56 -0500 (EST) Received: from localhost (teifel@localhost) by disney.csl.cornell.edu (8.11.3/8.9.2) with ESMTP id fADH2ox00133 for ; Tue, 13 Nov 2001 12:02:50 -0500 (EST) (envelope-from teifel@disney.csl.cornell.edu) X-Authentication-Warning: disney.csl.cornell.edu: teifel owned process doing -bs Date: Tue, 13 Nov 2001 12:02:50 -0500 (EST) From: "John R. Teifel" To: Subject: 615 PAPER 42 Message-ID: <20011113114357.J85469-100000@disney.csl.cornell.edu> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Tiny OS: Tiny OS is a very small event-driven operating system for use in embedded sensor networks. It is not really all that interesting. A fairly trivial application of standard embedded system O/S design. I am really unsure of any useful insight from this, other than that it is a viable building block. Smart Dust: This on the other hand, is a fairly novel idea and I am impressed at their ambitious goals in tackling the hardware integration problem inherent in very time sensor devices. Similar integration problems are being investigated in Cornell's ECE695 course this semester. They have a ways to go, however, to acheive the size that they are seeking. From daehyun@csl.cornell.edu Tue Nov 13 12:24:55 2001 Return-Path: Received: from wilkes.csl.cornell.edu (wilkes.csl.cornell.edu [132.236.71.69]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADHOrR19422 for ; Tue, 13 Nov 2001 12:24:54 -0500 (EST) Received: (from daehyun@localhost) by wilkes.csl.cornell.edu (8.9.3/8.9.2) id MAA50392 for egs@cs.cornell.edu; Tue, 13 Nov 2001 12:24:48 -0500 (EST) (envelope-from daehyun) From: Daehyun Kim Message-Id: <200111131724.MAA50392@wilkes.csl.cornell.edu> Subject: 615 PAPER 42 To: egs@CS.Cornell.EDU Date: Tue, 13 Nov 2001 12:24:48 -0500 (EST) X-Mailer: ELM [version 2.4ME+ PL54 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit TinyOS and Smart Dust are the system architecture for sensor networks. The technology advance such as MEMS, low-power communication devices enables the micro-level wireless sensor networks. However, each network element has still lots of limitations such as battery, communication bandwidth, memory and computation power. These two projects try to provide efficient architecture to overcome the limitations. In my opinion, the main problem of the micro-sensor networks is that they don't have good practical applications yet. Their micro-level size implies critical resource limitations, so provides good research topics. But, on the contrary, it can be the main obstacle that prevents these techniques from being used practically. And, in my opinion, it is still very hard to overcome those limitations with the techniques proposed here. From jcb35@cornell.edu Tue Nov 13 12:39:34 2001 Return-Path: Received: from travelers.mail.cornell.edu (travelers.mail.cornell.edu [132.236.56.13]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADHdXR21200 for ; Tue, 13 Nov 2001 12:39:33 -0500 (EST) Received: from travelers.mail.cornell.edu (travelers.mail.cornell.edu [132.236.56.13]) by travelers.mail.cornell.edu (8.9.3/8.9.3) with SMTP id MAA17935 for ; Tue, 13 Nov 2001 12:39:11 -0500 (EST) From: jcb35@cornell.edu Date: Tue, 13 Nov 2001 12:39:11 -0500 (EST) X-Sender: jcb35@travelers.mail.cornell.edu To: egs@CS.Cornell.EDU Subject: 615 PAPER 42 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Both of these projects at Berkeley deal with the design of wireless communication systems for distrbuting data over sensor networks. They focus on the design of small electronic systems that strive for longevity and functionality. Smartdust strives to show that a complete sensor network can be integrated into a cubic millimeter package. Each one of the "dusts" can communicate on some kind of wireless plane, whether it be using rf frequencies or other laser pointer devices. Using a laser device, the team was able to communicate over a 21km area (that had line of sight). The TinyOS group is an event based operating environment that supports threads and event driven state machines. They use a form of active messages to communicate between the nodes and establish a tree based on some sort of "base station" node. I thought they had some interesting applications, such as the standard sensor network to detect vehicles, which was cool because it was deployed by an unmanned air-vehicle. I also thought that the "keyboard glove" was an interesting application for a sensor network. The batter life for the sensors ranged from 72 hours (if it listened all the time) on up, depending on the amount of time they spent actually listening. I would be curious to see more traffic patterns and how they could be distributed so as to lengthen the longevity of the system. I would also be interested in seeing applications of the motes in sensor networks dealing with heterogeneous communication mediums. From samar@ece.cornell.edu Tue Nov 13 13:36:49 2001 Return-Path: Received: from memphis.ece.cornell.edu (memphis.ece.cornell.edu [128.84.81.8]) by sundial.cs.cornell.edu (8.11.3/8.11.3/M-3.7) with ESMTP id fADIamR28474 for ; Tue, 13 Nov 2001 13:36:48 -0500 (EST) Received: from aquinas.ee.cornell.edu (aquinas.ee.cornell.edu [128.84.236.57]) by memphis.ece.cornell.edu (8.11.6/8.11.2) with ESMTP id fADIZMH32618 for ; Tue, 13 Nov 2001 13:35:22 -0500 Date: Tue, 13 Nov 2001 13:35:41 -0500 (EST) From: Prince Samar X-Sender: samar@aquinas.ee.cornell.edu To: egs@CS.Cornell.EDU Subject: 615 PAPER 42 Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII 42) Sensors: Mote, TinyOS, Smart Dust Sensor networks are expected to be widely deployed in the future, owing to its low cost and small size. Some of the current sensors can sense acceleration, magnetic field, temperature, pressure, humidity, light and rf signal strength. The limiting features of sensor networks are the limited amount of power, computation, storage and bandwidth that they possess. This calls for really efficient designs that conserves the battery power and cost while keeping the size of the sensor small. Further the sensors are intended to be largely unattended and should be robust with narrow interfaces. TinyOS looks at this problem from the software point of view, whereas Smart Dust tries to optimize the hardware of the sensor. TinyOS is a lightweight multithreaded operating system for wireless sensor environments. They provide a high degree of concurrency to keep the data flowing and reduce the requirements for data buffers. TinyOS component model propagates hardware abstractions into software and allows for migrations of software component into hardware. The event based nature of TinyOS enables efficient use of the energy resources. One of the nodes is designated as the base station and all the rest of the nodes propagate data to this node. The use of Active messages, which is essentially a remote procedure call without the overhead of the system call, is an very interesting idea used by TinyOS to reduce energy consumption. Smart Dust are designed to be small sensors which can be spread in an area randomly and can monitor the area for long periods of time. Thus a key feature of the Smart Dust is the conservation of energy. The nodes are designed to have a very low duty cycle, waking up periodically to gather information and communicate with other nodes before going back to sleep again. They use an optical communication scheme, using reflectors to communicate with other nodes.