Anda di halaman 1dari 14

A Paper on

Speckled Computing
Submitted by
Bijo P Joy
Roll No 10BM60020
VINOD GUPTA SCHOOL OF MANAGEMENT
IIT KHARAGPUR

As a part of the course of Management Information Systems

Instructor: Dr. Prithwis Mukerjee

Abstract

Speckled Computing is a new concept in Information Technology that has the potential to
revolutionise the way we communicate and exchange information. This is a radical technology based
on minute autonomous specks, each of which encapsulates sensing, programmable computation and
wireless networking capabilities. Computing with minute specks will enable linkages between the
material and digital worlds with a finer degree of spatial and temporal resolution than hitherto
possible, which will be fundamental to truly ubiquitous computing. This will be achieved through the
computational aura that is created by the dense network of specs.

Introduction

The research on Speckled Computing undertaken by a consortium of four British


Universities is believed to revolutionize the existing world of sensors. Speckled computing
is a Nano-technology which is realized by minute, autonomous semiconductor specks –
each speck that encapsulates the capabilities for sensing, programmable computation, and
wireless networking. These multifunctional capabilities have enabled these Specks to act as
the final extension of the Internet.
The Specks can be sprayed into the atmosphere, onto surfaces or onto people, and
which would then collaborate to form programmable computational networks called
„SpeckNets‟ that paves way to the goal of truly ubiquitous computing. A dense and non-
static wireless network of thousands of these specks is called a Specknet – which extracts
information from the collected data.

The basic purpose of Specknets is to link the digital world of computers to the physical world
of sensory data. A Specknet on the person, for instance, is capable of tracking movements of the
limbs, or the position of the person in the environment, and this information can be stored,
manipulated and accessed remotely over the internet. Computing with specknets, or Speckled
Computing, affords new models of limitless interaction with the digital world, in which the physical
environment is the primary site of interaction. In short, Specks bridge the gap between digital and
virtual worlds.

Bringing down the three functions of computation, storage and communication into the „speck‟
level will bring about a complete change to the world of sensors. The current sensors that are
available in the market are big in size and often require an even bigger power source. Along with
these, they often need to be physically connected which leads to the added disadvantages for
deployment. With the introduction of Speckled Computing, studies and inferences can be reached
without altering the environment.

SPECKS

The specks are often in the dimension of very few millimetres. Their dimensions often vary in
the range 1mm*1mm*1mm to 5mm*5mm*5mm. The specks are made from the post- modern
semiconductor materials, which would enable it to hold it properties.

The specks are also non-static and unreliable. They are highly mobile. So there exists a
problem in using specks individually. The specks also have limited power supply. The major aim is to
develop a system that would provide the sensory capabilities of the specks coupled with the long
lasting power supply. Computing with minute specks will enable linkages between the material and
digital worlds. This will also be with a much finer degree of spatial and temporal resolution than
hitherto possible.

The three major tasks with the specks are:


- Sensing
- Processing
- Communicating
There are certain design constraints related to the development and manufacture of these
minute, autonomous specks. The specks have:

 Limited power supply


 Limited resources for energy scavenging
 Limited memory

The major constraint with the specks is that of its power requirement. The main goal is, thus
to reduce the power requirement as much as possible. The researches have come up with specks
that will work in different modes. It can work in sense and memory mode as well as remote memory
mode. In the sense and memory mode, only the sense and the memory units are active. In the remote
memory mode, only the memory and communication units are active. This helps in saving quite a
considerable amount of energy. The transfer in these operation modes occur as a response to the
events from either the program world or the outside world.

1. Speck placed on a finger (Source : www.specknet.org)


2. Speck : A Close-up (Source : www.specknet.org)

SPECKNET

A dense and non-static wireless network of thousands of these specks is called a Specknet -
which will collaborate to extract information from the data. Specknets are formed based on a property
of the specks. The specks can communicate wirelessly over a few centimetres. So when kept close
together, they form a dense and wireless network of thousands of specks. Specknets link the digital
world to the physical world. The speck on itself is quite useless. This is because they do not have
enough memory space and processing power to perform the computations that will be required in
practical applications.

The specks only have a very small amount of memory, in the range of very few kilobytes. This
would allow us to run only very simple programs in the specks. It is also impossible to route a
connection between thousands of these specks with the current memory constraint. An individual
speck moves around a lot and hence is very unreliable. It also spends a lot of time sleeping. So the
better way is to consider the Specknet rather than a single speck.

Specknets are to be managed properly. Managing these specknets is a laborious task. To


make it simple we divide the Specknet into many Speckzones. Group of specks of approximately the
same size are the speckzone. Proper mechanism has to be maintained to make sure that a speck can
only be part of a single Speckzone. The high mobility of Specks makes this all the more difficult.
The concept of maintaining Speckzones rather than individual Specks is highly beneficial. It is
easier to program Speckzones as a whole than to program specks. This provides better memory and
processing power. By introducing the concept of treating the specks as a group called the Speckzone,
the routing becomes much better. It is highly time consuming to route individual specks. The use of
the Speckzones makes it far more reliable as compared to the individual specks.

The speckzone must continue to work even if some of the specks drift away or run out of
power. The size of the speckzone is an important constraint; the size of the speckzone must be
managed within a size that can be managed. This is owing to the highly mobile and unreliable nature
of the specks.

The algorithm for this must be chosen and implemented carefully. It must be in accordance
with the decentralized and leader-less nature of the specknets. Hence the algorithm in the
Speckzones must be decentralized and leader-less. It is also to be made sure that the algorithm uses
only local information for computations.

The size of the Speckzones must be optimal. This is because of the fact that even though,
large Speckzones have high processing power and memory, and it might be difficult to manage the
Speckzones because of the large size. The processing power is still, by large a major factor that
might affect the performance of the specks, the Speckzone and also the Specknet.

It should be also taken care that the specks are placed close enough to each other. This is
because of the fact that, power requirement of the specks for inter-zone communication is large if the
specks are too far from each other. So the specks are to be hand-placed and are to be kept as close
as possible.

The network of the specks, the Specknet, forms a computational aura in the field. The
speciality of the computational aura is that, it enables the accurate determination of the finest details
of the processes that happen within the aura. The computational aura is formed by suspending
individual specks in the field or by spraying the specks into the atmosphere.

Specknets provide a means of distributed computing. The distributed computing is purely a


fine-grained computing system. This is realized by the numerous specks that contribute to the
computing power. The entire Specknet can be considered as a distributed computing system. This
paves a way for the concept of ubiquitous computing.

Ubiquitous computing is generally referred to as the third wave in computing. The first wave
was the concept of workstations where in many people shared the same computer. The second wave
comprises of the personal computers where a person uses a single computer for computation. The
third wave of the computing is the ubiquitous computing where a person has many computers for
computation.

According to Mark Weiser, the father of Ubiquitous computing, it is roughly the opposite of
virtual reality. Where virtual reality puts people inside a computer-generated world, ubiquitous
computing forces the computer to live out here in the world with people. Virtual reality is primarily a
horse power problem; ubiquitous computing is a very difficult integration of human factors, computer
science, engineering, and social sciences.

COMPARING SENSOR NETWORKS AND SPECKNETS

The specknets needs to be compared with the existing sensor networks to understand the
advantages. The existing sensor networks are way behind when we consider different factors that
affect the computational and the sensing power of the networks. The network of the specks, the
specknets are all program –centric where as the existing sensor networks are all data-centric.

The existing sensor networks are sparse, that is the density of the sensors is very low. But the
specknets are formed by closely packing the specks so as to reduce the power consumptions of the
speckzone. The specknets are thus dense. The nodes (sensors) in the sensor networks are all
immobile. The stay in a place and acquire data or information. Whereas the specknets are mobile due
to the mobile nature of the specks. They gather information by continually travelling in and around the
speckzone.

In the case of the sensor networks, data-transfer model is used. is entrusted with the duty of
transferring data from source node to the sink node. In the case of the specknets, the data transfer
model used is peer to peer. Data transfer occurs by considering each of these nodes at the same
hierarchy.

The control model for the specknets is leaderless and decentralized. There is no special right
given to any node. All nodes are considered the same and are given the same priority. The
communication occurs through any of these closely lying specks. The network thus formed is efficient
and proper communication takes through these specknets.

Specknets can act as an extension of the internet. It can be treated as the last millimetre of
the internet. The finite data that is determined from the computational aura can be send through the
internet. The internet can be connected to the Specknet which would act as the end-systems in the
internet, which is given the job of a sensor.

This technology is of immense applications. This very feature enables the use of specknets in
the field of medicine and physiotherapy. The specks sprayed on to the body of a patient forms a
network among themselves and then is entrusted with the duty of sensing pressure and temperature
variations in the field that is formed due to the presence of the specks.

DETERMINING THE POSITION OF THE SPECKS

One of the biggest challenges is to accurately determine the position of the Specks when they
are performing the functions of a sensor. This is achieved by introducing certain checkpoints in the
Specknet known as the anchor nodes. The anchor nodes are associated with the hardware for
determining the position of the specks. They are much bigger in size as compared to the Speck.

3 Specknet : A depiction (source : www.specknet.org)

These anchor nodes periodically broadcast their position. These anchor nodes knows their
position all the time. Thus the anchor nodes can be used to compute the position of other specks. The
specks calculate their position with reference to the anchor nodes that are placed throughout within
the Specknet.

The specks can determine their position as an average of the distance from all the anchor
nodes, which they had received over a period of time. Calculation of position by such means restricts
the specks to take only the distance information that they had received over a certain interval of time.

Anchor nodes are to be placed carefully with the atmost attention. Anchor nodes are to hand-
placed, fixed in position and then individually told where they are. This has to be done properly, or
else they have adverse effects on the position determination of the entire specks.
The anchor nodes do have additional hardware associated with them. This mainly includes a
Global Positioning System (GPS) –like software. The anchor nodes are also to be evenly distributed.
This would make an inter-leaved, overlapping effect on the field of the anchor nodes.

INTER-SPECK COMMUNICATION

The major attraction in the case of Specknets is the ability to communicate between the
specks. If the inter-speck communication is not proper, the entire system will be faulty. The
communication among the specks can be achieved with the help of a gradient system, which uses a
referential mode for communication between the specks.

In the gradient wave method, the source of the message in a communication will emit the
message along with the counter 0. The specks receiving these messages shall accept the message
and use this for its different computations. The message is further retransmitted after incrementing the
counter by one. Thus, the system allows the recipient node to know where exactly the message was
originated and how far the source is.

Usually the messages with the higher count values are all neglected. The messages with the
smaller counter values are accepted and retransmitted while the others are accepted and neglected
later. The advantage with this technology is that, the specks can always determine the messages that
are out-dated and can thus opt for the latest updates. This also provides an overview regarding the
nodes through which it has travelled to reach the destination. The method has proved to be better
than all the other techniques that were tried during the passing of messages between specks.

PROGRAMMABLE SPECKS

The specks are programmed with the needed software. As a part of research, a series of
programmable specks were released. They are called the Prospeckz. These specks contain a
processor, a radio and a PSoC. Three versions of Prospeckz have been released - Prospeckz I,
Prospeckz II, Prospeckz III.

The specifications of the Prospeckz III are given:

 Atmel AT91FR40162 processor


o 60 MHz ARM7TDMI core
o 256 kB RAM, 2 MB Flash
 ChipCon CC2420 2.4 GHz ZigBee radio

 Cypress CY8C27643 PSoC


o 256 bytes RAM, 16kB Flash
o configurable analog circuitry
o sensors: light, sound, temperature

4: Components in a Speck (source : www.specknet.org)

THE ARCHITECTURE

The architecture is comprised of three layers. They are the speck architecture, the Specknet
architecture and the speckled computing layers. The specks include a group of sensors, which does
the sensing function. These sensors are a part of the PsoC that has been embedded in the speck.
The processor does the computation of the problem at hand. There is also an energy source that
provides the power to work. The energy source is to be so chosen that the speck is supplied energy
as long as possible.

The radio is a ZigBee radio, which along with a photonics cell that does the job of networking
and the physical layer communications that deals with the communication between each of these
units. The co-ordination between each of these components is to be kept since the speck forms the
base of the entire concept of speckled computing. The management of the speck-components cannot
be done easily, owing to the nanometre dimension of the various components. So the architecture is
so as to reduce the damage as far as possible.

Above the speck level architecture, we have the Specknet. The Specknet level corresponds
to a higher level of abstraction of dealing with specks. Here we deal with Speckzones and Specknet
as a whole rather than specks. The Speckzones act as a harness of computers which does the
computations. This could be taken as a case of distributed computing wherein each of the speckzone
is acting as an individual system.

5. Speckled computing : Architecture (source : www.specknet.org)

At the higher level, we have the speckled computing wherein the demonstrators and the
programming environment are the key players. The speckzone which has been programmed creates
an environment or an aura within which any event can be accurately determined and the results (local
data) which can be used for further computations.

PROTOCOLS

The Specknet architecture includes the group of specks (speck zones) that form the Specknet
and the communication protocols that exists among them. There exists the need for a routing protocol
that can transmit packet to nodes across a network and also be proactive and reactive. A protocol is
basically a set of mutually agreed rules that is to be kept during the course of a communication.
Proactive algorithms require large network capacity for the routing table maintenance.
Distance Vector algorithms are such kind of algorithms. Reactive algorithms have long route
construction delays. Ad hoc - On Demand Vector routing (AODV).

Studies have shown that the Destination Sequenced Distance vector routing is suitable for the
Speck-based communication. There was no considerable dip in the number of packets dropped. Also
the Tx/Rx time was also comparatively very low.

The Specknet can render existing MANET and sensor network protocol (DSDV), but the
memory constraint has to be kept in check. A novel protocol- Speckzone based routing protocol
(SBaRP) enables data to route data under stringent memory requirements. It is a hybrid routing
algorithm that combines the advantages of both proactive and reactive algorithm to enable routing in
specknets. The speckled computing layer includes the programming environment and the
demonstrators.

APPLICATIONS

The application realm of the technology of specks and speckled computing is quite large. It
comprises of Asset or People Tracking, Environmental and Industrial Monitoring, Health and Safety,
Security, Games, Human computer Interfaces etc.

The concept of speckled computing will revolutionise the tracking of people and things. The
high end and accurate data that the specks provide can be used for precise determination of the
location and the variations of the subject. For example, the specks sprayed onto the body of a
dancer, will provide such a system. The specks would create a Specknet among themselves and
then keep a check and sense the variations that occur in him/her.
6. Processed image of specks used to track the movements of a Dancer (source: www.specknet.org)

The data includes the speed of rotation, movement, temperature, pressure, energy spend etc.
This provides a very useful service at disposal. Also the same technology can be used in the medical
field too. The specks sprayed onto a patient can track the exact condition of the patient and thus
helps in diagnosis. There is an added advantage to this. The Specknet formed could be connected to
the internet and thus the data can be send over the net. The same technique can be used in an
industry for the asset tracking.

The technology of specknets can be used in the Human-Computer Interfaces. The HCI would
requires Position, distance, velocity or acceleration sensing, other forms of input like the audio, light
etc., Output/actuators and Displays that are flexible, cheap, rugged and embedded (Shift away from
projection). The technology would pave the way for cheap HCIs. The tangible interface is another
application.

Yet another application of the speckled computing is in the urban-location based application.
It deals with the Creation and Visualisation of Multiple “Situated Narratives”. i.e., it Captures city wide
activities. The system provides details of the habits and living style of the people of a city. In the
normal case, it is being realise with the help of cameras. With the advent of specknets, it is now
possible to create specknets which would communicate with each other and thus enables us to
create a virtual model of the entire city, which in turn provides an overview about the city.
The specks and speckled computing has got immense applications in the field of gaming. It
could be improvised to be used in the game play. A headset which is embedded with Specks can be
developed so as to pick up the electromagnetic pulses emitted by the human brain, which are also
known as the brain waves. This would be a mini Electro encephalogram (EEG) which in effect that
picks up stimuli from the sudden fluctuations in the thought process and helps in game play. The
same technology can be used to simulate war situations for training the armed forces, thereby paving
way for a cheaper and much effective preparation.

The applications just don‟t end there. The Specks are so convenient that even the tennis
racket of a player can be embedded with it. The power, speed of movement etc. can be monitored
from a different location to provide even the minute details. Similarly, the day when a cricket ball or a
Football or a puck embedded with Speck sensors are not far.

In short:

 Specknet is timely and fundamental technological advance


 Specknet is an excellent match to recent advances in HCI
 Specknet will enable further paradigm shifts in HCI and novel applications
 Specknet – A powerful combination of technological development and application domain
expertise

CONCLUSION

In this age of calm technology, when technology recedes into the background of our lives, the
concept of Speckled computing play an important role. Specknets link the digital world of computers
to the physical world of sensory data. Computing with specknets, or Speckled Computing, affords new
models of unencumbered interaction with the digital world, in which the physical environment is the
primary site of interaction. Specks bridge the gap between digital and virtual worlds.

The application area of the specknets is quite large. They will revolutionize the fields of asset
or people tracking, environmental and industrial monitoring, Health and safety, security, games etc.
The application realms of the speckled computing are vast. They mark a new era in the field of
computing. In short, Speckled Computing will surely change the world, the way we look at it.
REFERENCES

1. The website www.specknet.org , which is the speckled computing website and is used for the
co-ordination of the researches among the four different British Universities.
2. Reardon, C and Krauss, T, "Specknet: Optical Networking and Location Discovery",
submitted to IEEE Pervasive Computing, 2006.(Journal)
3. Elgaid K, McLelland H, Ferguson S, Zhou H, Wilkinson C D W, Stanley C R, Thayne I G, “On
the viability of mm-wave array-based MMIC design using 50 nm InP HEMTs and room
temperature deposited SiN MIM capacitors”, submitted to IEEE Transactions Electron
Devices. (Journal)
4. Elgaid K, McLelland H, Holland M, Moran D A J, Stanley C R, Thayne I G, “50 nm T-gate
metamorphic GaAs HEMTs with fT of 440 GHz and noise figure of 0.7 dB at 26 GHz”,
accepted for publication in IEEE Electron Device Letters, 26, 11, pp 784-787, 2005. (Journal)
5. Elgaid K, Zhou H, Wilkinson C D W, Thayne I G, “Low temperature high density Si 3N4MIM
capacitor technology for MMMIC and RF-MEMs applications”, Microelectronic Engineering,
73-74, pp452-455, 2004. (Journal)
6. Davoudani, D, Hart, E and Paechter, B, "Computing the State of Specknets: Further Analysis
of an Immune-Inspired Model." In Artificial Immune Systems, Proc. 7th International
Conference, ICARIS 2008, Phuket, Thailand, August 2008, vol. 5132, pp. 95-106. Springer,
2008.(Paper presentation)
7. McNally R and Arvind D K, “A Distributed Leaderless Algorithm for Location Discovery in
Specknets”, Proc. Euro-Par 2007, Rennes, France, 28-31 August 2007, Springer-Verlag.
(Paper presentation)
8. Darbari F, McGregor I, Whyte G, Stewart R W and Thayne I, “Channel Estimation for Short
Range Wireless Sensor Network,” pp 25/1 – 25/8, 2nd IEE/Eurasip Conference on DSP
Enabled Radio, UK, September 2005. (Paper presentation)

DISCLAIMER

The write-up is based on a technology which is currently a research undertaken by the


consortium of four British Universities. The article is completely based on the author‟s interpretation of
the data available at www.specnet.org and its possible future applications. The article is not meant to
be a reference. Its purpose is to introduce the concept and the author shall not be responsible if any
person uses it for any other. No person or entity is allowed to use the pictures, graphs or data without
prior permission from the concerned authorities.

The Author

Anda mungkin juga menyukai