Academic review - ANU Edge

beadkennelΤεχνίτη Νοημοσύνη και Ρομποτική

15 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

97 εμφανίσεις


ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

ICT landscape

A REVIEW OF THE GROW
TH AREAS IN ICT

ACADEMIC REVIEW

Thank you for doing an interview with Mathew for the ICT landscape to give Lockheed a better
idea of
the current trends in ICT
. Below is a table that contains summaries of

the top forty
fastest growing research areas. This list was informed by
inform
ation
taken from the interviews,
supplemented by a

world
-
wide
bibliometric analysis

and
a brief
literature review
.

The table is not a list of the expertise of the ANU; it is intended as a broad snapshot of the big
growth areas of ICT research around the world.
The descriptions are meant to be broad
,
but
concise.
Lockheed will use this list to identify which areas they would have most interest
working in.

Could you please scan the list of 40 and for each item that overlaps with your research, check
the descripti
on to ensure we are covering the major points.
Then answer the following three
questions.

Which ones overlap with your research?


Which five of these growth areas will create the biggest global impact in the next ten years
?


Is there anything that is missing that you think should be covered.




Please return to
mathew.mcgann@anuedge.com

by
Midday Wednesday 19
th

June
.

Thank you.





ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Hardware and C
omputation

Research

Growth
Areas

Description

Heterogeneous
architecture

Limitations on computational power have shifted priorities from a do
-
it
-
all processor in favour of a network of specialised processors.
These new processors require special low level software. By
programming th
e communication between processors, the hardware
itself becomes "programmable". Field Programmable Gate Arrays
are an example.

High power requirements
of HPC

High performance computing is needing greater and greater amount
of energy

leading to increase costs and environmental impact. The
complexity of some proposed
simulations

could incur cost tens of
thousands of dollars. This leads to research into the reduction of
energy requirements for computing both through hardware and low
leve
l software optimisation.

Massively multicore
computation

High powered computing is making progress by adding more and
more processors and dividing the computing between them. At
current rates of progress, exaflops are expected in ten years,
provided some

of the challenges of moving to a massively multicore
system can be solves. Problems to solve include increasing the
connection speeds between processors and software to best
manage this "sparse" hardware.

Distributed (grid)
computing

The distribution of

a computational task across multiple devices has
been done for some time. But as networks get larger and more and
different devices are connected, the potential of distributed
computing has grown. Such diversity in connection, devices and
operating system
s make new generations "grid" computing systems
growing research area.



ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Human
-
Computer Interaction

/ Computer Vision

Research

Growth Areas

Description

Ubiquitous seamless
computing

The popularity of home computers followed by laptops, then
smartphones to tablets and perhaps to Google Glass marks a trend
where computers are becoming less of a tool and more of a part of
our everyday experience. These new devices are expected to be
ever
ywhere (ubiquitous) and applications need to ensure they can
be experienced on all devices, so that the transition between them is
seamless.

Human
-
Technology
interaction

The rapid growth in technology integration into society fosters
research into how te
chnology affects humanity and society. This
area contains research in behavioural IT, marketing (how to make
people comfortable with IT), legal concerns (who is responsible for
coding errors or bugs), and regulation (for example: regulation of the
internet
). There is also interest in the behavioural responses to
technology and innovation including the "luddite effect", why people
may disengage with technology.

Human
-
Computer
Interaction (HCI)

How computers detect humans is covered in "computer vision", th
e
problem of how computers understand humans is another matter.
Making new computer systems "human friendly" is important,
especially in machine learning, where it is important for the computer
to understand the user's intent, and perhaps even predict what

the
user wants in advance.

Biometric identification

There is active research in multiple methods of uniquely identifying
people from different body parts. Fingerprint and palmprint, veins,
irises, faces, ears. Such identification systems have security
applications, but when applied to computer vision can extend to
more advanced emotion identification, which may also advance
human technology interaction.

Computer vision

Computer vision is a growing research area that aims to make
computers capable of
i
nterpreting

scenes identifying what they
represent, what they contain, etc. This has applications in
autonomous navigation (road imaging), video searching (scene
recognition), medical diagnoses (cancer imaging), human
interpretation (Xbox Kinect is an exam
ple) and much more.

Object tracking

An aspect of computer vision, object tracking involves the additional
task of not only detecting objects, but following them through time.
And keeping track of them if they temporarily go out of view. This
area has app
lications in automated surveillance. The temporal
resolution also allows prediction, and now behaviours may be
monitored and fit to known behaviours.


ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Position monitoring

GPS systems are very common, but only work outside when the sky
is visible, therefore there is active research broadening the
environments for applying this technology. Not just improving the
resolution of the satellites systems, but also providing localis
ed
indoor positioning. Cheap GPS devices or good devices with poor
reception can be combined with other sensors (e.g. inertial sensors
and on
-
the
-
ground imagery) to improve localisation.






Big data

Research

Growth Areas

Description

Decision support

Decision management systems combine multiple sources of data to
provide information, hypothetical extrapolations or probabilistic
predictions to guide decisions in the real world. When combined with
real
-
time sensor networks real time decision support can

be
realised. A major application would be responses to disasters and
disaster management.

Policy testing and
complex system
intervention

The use of big data in social sciences, biology, and the humanities
has potential to model the effectiveness of policies or specific plans.
Doing so in real time with an always
-
on data stream allows progress
of the plan to be monitored and will enable effe
ctive intervention.
Research is ongoing as to how best to deal with real time
information, and how to test intervention against controls.

Data type and storage
for big data

The major challenge in creating helpful analyses of big data is
combing the data
into one place. This includes two problems,
difficulties around the storage of the data (up to petabytes and
beyond) and its format (including NoSQL databases). The format
problem is particularly acute as existing useful data is diverse (in
different, inco
mpatible formats) and new data is dynamic (rapidly
added and changing)

Anonymous data mining

Privacy is an important aspect in big data mining, with more people's
credit ratings, and civic records being compiled and searched, the
privacy of these individu
als is difficult to ensure. However there is a
growing field of anonymous data mining which combines
cryptography and access techniques so that information that
identifies individuals is not known to the software or investigator.



ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Machine Learning, Ontol
ogies, Semantics

Research

Growth Areas

Description

Optimisation

Applications of optimisation include making more
efficient

use of
infrastructure, optimising ren
ewable energy sources, determining

the
best navigation route, and many more. Thus study into techniques
of optimisation is a very large research area. Some methods are
inspired by nature, for instance swarm optimisation, which is a
particularly strong research area.

Machine learning

Sys
tems that learn from new data have potential to improve
performance without the programming being modified. Machine
learning broadly allows inference and analytics on datasets that
generally improves over time, reflecting the idea that machine
learning rep
resents statistics for the 21st century. Applications are
endless, but include text analysis and methods of computer vision.

Knowledge
representation

Computers are increasingly able to do tasks only humans were
capable of, resulting in the automation of knowledge work. Digital
personal assistants could be capable of intelligently organising work
and solving problems. Research in this area revolves aroun
d how to
represent facts and how logic works on these facts to infer new
knowledge elements.

Semantic web and
searching

A growing area of research is the improvement of search engines by
including a semantic capability so that the engine understands the
intent of the user. Applying such searches to big data, semantics
has the potential to improve searching, and can provide probabilistic
queries. This includes understanding the context of searchers and,
more generally, defining ontologies: shared vocabular
ies to allow
reasoning about concepts.



ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Web 3.0

Research

Growth Areas

Description

Personalised experience

Recommendation engines that personalised people's online
experiences are common, but as computers become ubiquitous and
as sensor networks record more data, other aspects of life are
expected to be personalised including a person's medicine and their
exper
ience with their government. Recommendation and filtering
systems are a growing area of research, as are consumer trust
modelling.

Quality of Service

The monitoring and optimisation of the quality of communication
methods, computer networks, and online services is a growing area
of research. Quality of service in communication and computer
networks overlap with the transmission research area. Online
ser
vices, such as those in the cloud, need to distribute uptime
between many users while maintaining their system performance.

Hybrid cloud

As cloud services become more common, new problems have
arisen. How can multiple cloud services work together? How ca
n
different services to best brokered? How will big data work over the
internet. There is growing research in this field answering these
questions.

Analytics for social
networks

There is currently a growing research interest in social networks.
Potential

capabilities arising out of this include the methods to better
visualise and understand complex social networks, the
determination of online cultural boundaries and the ability to predict
how viral a message may be.

Games

There is growing research inter
est in games. Both in artificial
intelligence (how can computers be coded/trained to play games)
and how humans play games. Massively Multiplayer Online games
have developed worlds in which feedback and rewards can be
tweaked and the results monitored in r
eal time
-

systems have
developed to optimise people's performance by turning tasks into
games and optimising their rewards (gamification).



ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

ICT in real world

Research

Growth Areas

Description

Tele
-
education

Technology is increasingly being distributed among schools, and the
ways to utilise the technologies and which kind of software would be
most effective is a current area of research. At the tertiary level,
universities are opening to provide entire courses

online, teaching
remotely with interactive web
-
based software.

Telemedicine

The majority of doctor time is spent diagnosing minor conditions, in
some cases the diagnosis process is simple enough for a computer
to solve. Large hospital data sets contain
information to make
diagnoses more accurate, such systems can allow control groups to
be more easily monitored for experiments and online video or survey
interaction can replace face
-
to
-
face treatment saving time and
resources. Current research is tied to
big data research, and the
technology is independently being applied in hospitals.

IT applied to medical
experimentation

ICT is increasingly used to aid in medical research, including
modelling of biological phenomenon (e.g. blood flow in arteries),
determining health
-
lifestyle correlations, modelling epidemics and
epidemic interventions, determining gene expression and moni
toring
and measuring an individual's health. This area overlaps with
telemedicine as ICT also helps optimise and run hospitals.

IT applied to
manufacturing

Manufacturing is using ICT to improve products, quality and
methods of production. Asset managemen
t systems can manage
supply chains and logistics, and control inventory. Related to this is
the idea of lifecycle monitoring, so maintenance can be scheduled
and problems pre
-
empted for minimal downtime. On the production
side, sensors and software are bei
ng used to test material and
device quality. Systems are also being developed to monitor
production and machine performance, which provides a powerful
overlap with resilient systems.

Massive science
experiments

The data produced by large, complex science experiments such as
the Square Kilometre Array, or online protein sequencing
experiments are textbook examples of big data. Researchers are
increasingly interested in how to store and access large datasets.
New e
xperiments may be avoided if similar results have already
been recorded. In these systems, storage will be a service as much
as simulations are.

Transport optimisation
and automation

Research into traffic monitoring, modelling and optimisation has
never
been more popular, and the available technology is such that
these systems are finally likely to be adopted. This area also
includes research into navigation, and near and complete vehicle
automation.


ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Networks, Sensors and Control

Research

Growth Areas

Description

Virtual and conventional
computer networks

Research into computer networks covers the sharing of unstructured
data and internet based developments like improved streaming.
There is strong research activity in virtual machines and networks of
such machines. In particular there is demand for scalable and
efficient networks, as more services move to the cloud.

Wireless networks

The use of wireless networks is growing. There is still research in
how to improve such networks. The pervasiveness of
mobile
networks has other important implications, including security
(integrity of mobile data, difficulty of tracking mobile sources) and
work mobility.

Ad
-
hoc networks


There is significant attention on wireless ad
-
hoc networks that do
not rely on infrastructure. Devices as nodes relay data to other
devices, cooperating to produce a connected network. This is
expected to lead to the ability to generate spontaneous, mobil
e
networks. These have application in situations where no network
infrastructure exists, e.g. remote areas,
military

deployment
applications and developing countries.

Sensor networks

Research accessing large sensor networks often in real time to
capture a
ccurate and time consistent information. Large scale
sensors may include geological sensors or remote autonomous
surveying drones and could be connected by ad
-
hoc networks or
may be built upon existing large networks (highways, 3G, NBN, the
Internet, etc.)
. On the smaller scale sensor networks can monitor
brain activity or people's behaviour in a smart home. As sensors
become more common and connected to the internet, the internet is
expected to become more of an "Internet of Things".

Transmission

Transmission includes research into techniques to control the
performance of communications across wireless, wired and optical
networks, both to improve the reliability, stability and capacity of
carried signals; or to inhibit or delay unwanted transmissio
ns.
Current

research focuses on radio spectrums and channels,
including dynamic (or cognitive) spectrum management and
cooperative frequency management, and reaches into applications
like sensing.

Resilient systems and
software

The field of resilient systems is an interdisciplinary approach to
control complex systems. Control is achieved with a combination of
human and automated intervention, informed by real time data.
Current research seeks to develop systems that are resilient

under
parallelism and complexity, to control feedback and to understand
nonlinearities that can make systems unpredictable.



ICT LANDSCAPE

A REVIEW OF THE GROW
TH AREAS IN ICT

Security

and Data Protection

Research

Growth Areas

Description

Copyright, watermarking

There is growing desire to be able to identify the ownership of a
given digital signal. This can be done by hiding (using stenographic
techniques) some marker in the signal so that it is undetectable
except under special conditions, this is "watermarking".

It can also
be used to verify authenticity or hide messages or can be used as a
mark of quality.

Cybersecurity and
conventional security

As more and more personal data is captured through devices such
as smartphones, the liability of this data and the protection of privacy
has become a serious concern. At the same time new ICT
technologies enable crime identification with artificial intelli
gent
programs able to track the MO of offenders and "bad
neighbourhoods" of ISP. While credit checks can be more accurate
and in real time.

Quantum cryptography

Quantum cryptographic systems are starting to be used in the form
of quantum key distribution
s. Other forms of quantum cryptographic
systems are less developed, but as a whole cryptography is the
most developed aspect of quantum computing, the basic utilisation
of which is 20
-
25 years away.

Formal Verification

Formal verification can prove the
correctness of an algorithm.
Currently, proofs of electronic logic circuits are possible and
research is now extending to verify the function of software.