Research Challenges in Networked Systems

The 5th VERDIKT Conference Research Challenges in Networked Systems

Torsten BraunResearch Group “Communications and Distributed Systems”Institute of Computer Science and Applied MathematicsUniversität Bern

Torsten Braun: Research Challenges in Networked Systems

2

Overview

> Recommendations from Evaluation of ICT Research in Norway> Looking Back: Topics in the Year 2000> Future Research Topics and Challenges> Remarks on Research Organization> ICT in EU Horizon 2020> Conclusions

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

3

Statements about the Future I

> It is difficult to make predictions, especially about the future.(Karl Valentin, Mark Twain, Winston Churchill, Albert Einstein, Niels Bohr etc.)

> Most technologies that will be used widely in 5-10 years are already existing in a lab today. We just do not know which of them will be successful. (Paul Kühn, KiVS Conference 1999)

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

4

Evaluation of ICT Research in Norway

Recommendations:“Research areas are often fragmented and not well aligned with Norwegian industry. A bottom-up «free academic» approach is dominating …. Although the outcome of this funding is not bad, coordination is lacking …. An obvious strategic area is ICT in life sciences, but also industrial robotics in dangerous environments such as off-shore, and cyber‐security. ICT in Energy and Power Engineering also seems also an obvious strategic research area of national interest.”

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

5

Networked Systems

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

6

Looking Back: ToCs in the Year 2000

> Wireless sensor networks and multi-sensor data fusion

> Denial of service and intrusion detection

> Internet-connected robots> Ubiquitous computing> Mobile agents> E-commerce> Digital libraries> Web services> Multimedia over ATM

> Software defined radio> Programmable networks> Voice over IP> Mobile ad-hoc networks> Beyond 3G mobile networks

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

7

Potential Future Research Areas in Networked Systems

25. April 2012

Energy-Efficiency

CloudComputing

Software-Defined

Communi-cations

CognitiveNetworks

SmartGrids

Cyber-PhysicalSystems

Partici-patory/

Opportunistic Sensing

FederatedSocial

Networks

FutureInternet

Security

Torsten Braun: Research Challenges in Networked Systems

8

Energy Consumption by Communications and Computing

> ICT already consumes approximately 2 % of energy, which is similar to air traffic. Trend is increasing.

> Approximately 10 % of ICT energy consumption by mobile communications (mainly base stations and mobile devices)

> LTE is even more energy consuming due to complex signal processing (MIMO, OFDM, etc.) than 3G.

> Trade-off between performance and energy consumption (→ optimization using power models, heuristics, and analytical techniques)

25. April 2012

9

Energy-Efficient Communications and Computing

> More energy-efficient hardware, e.g., Ps, storage

> Protocols and mechanisms (e.g., packet scheduling and aggregation, disruption-tolerant networks) to switch off systems / components or enable low power states; smaller cells, relaying, directed transmissions, etc.

> Energy-aware systems> Cross-layer mechanisms from

applications to physical layer> Cloud computing

(communication vs. computing)> Distributed processing and

process migration 25. April 2012

Torsten Braun: Research Challenges in Networked Systems

Torsten Braun: Research Challenges in Networked Systems

10

Cloud Computing: Problems

> Long delay between users and cloud computing infrastructure> Cloud computing assumes permanently connected (mobile)

devices and heavy communications. > Processing and storage of sensitive data> Availability and robustness > High energy costs for data centres, in particular cooling

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

11

Cloud Computing: Challenges

> Use data centres close to user, data might follow mobile users> Virtualization / isolation of computer and network resources> Disruption-tolerant mechanisms and protocols

for cloud computing> Privacy and trust mechanisms> Protection against

denial-of-service attacks> Replication vs. consistency> Green data centres > …

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

12

Software-Defined Communications

More and more functions previously implemented in hardware or special-purpose devices will be implemented in software on general-purpose computers, e.g., > Software-Defined Radio

— High flexibility in terms of frequencies and protocols, cf. cognitive radio/networks, although energy consumption could be an issue

— Future wireless systems might process even physical and medium access layers in software

— Signal and protocol processing in the cloud allow global mobile telecommunication networks.

> Software-Defined Networks— control network devices (routers, switches, access points)

via open APIs from external control plane software, e.g., OpenFlow— could simplify network configuration, e.g., in home networks

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

13

Cognitive Networks

> Cognitive Radio— Flexible adaptation to unused frequencies to better exploit

the frequency spectrum (higher bandwidth and reliability)> Generalization → Cognitive Networks

— Application of cognitive radio approach to higher protocol layers

— Full parameterization of protocols based on learning and intelligent algorithms

— Generation of tailored / customized software based on protocol descriptions

25. April 2012

Observe

Orient

Decide

Act

Learn

Environment

Torsten Braun: Research Challenges in Networked Systems

14

Smart Grids

> Automatic and distributed monitoring and control/adaptation of energy consumption, production, storage, distribution, pricing based on — accurate usage measurements by metering devices / sensors — interconnection of all components of the electricity grid

(sensors, control elements, actors)> Most electricity networks have

not been designed for asymmetric and varying flow of energy and information.

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

15

Smart Grids: Challenges

> Intelligent, distributed, automated control and coordination possibly based on processing large amount of data, learning, and prediction

> Heterogeneous, adaptive, reliable, secure, self-managing networks

> Secure systems and networks (privacy, authentication, intrusion detection, etc.)

> …

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

16

Cyber-Physical Systems

> Control / Computing Co-Design> actor/sensor networks & robots

integrated into physical world> have typically

real-time requirements> might use distributed /

high-performance / cloud computing infrastructures

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

17

Cyber-Physical Systems: Application Examples

> Critical infrastructures: power grid, transportation

> Traffic management> Manufacturing> Teams of (mobile)

autonomous robots> Building control> Medical care

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

18

Cyber-Physical Systems: Challenges

> Distributed real-time systems and control processing> Robustness> Communications and interoperability between

systems and devices > Security of systems and devices

(confidentiality, data and service integrity), in particular to prevent attacks on control systems

> Machine-to-machine communication (→ security)> Mobile CPS> …

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

19

Participatory and Opportunistic Sensing

> Motivation: Mobile end systems, cars etc. with several sensors> Approach: Exploit

— sensors, storage and processing capabilities of mobile end systems— mobility of users

for sensing instead of deploying huge amount of static sensors. > Application Examples

— Environmental monitoring— Traffic monitoring

and management> Challenges

— Feature extraction and data fusion

— Reliability, authenticity and accuracy of data

— Privacy— Incentives for and

coordination of users— Relations to social networking, e.g., tagging of events

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

20

Federated Social Networks

> Today: centralized social network architectures— Social network provider knows all data about users and their relations— Disadvantages of central solutions, e.g., robustness, security

> Tomorrow (?): distributed/federated social network architecture require— Open protocols and APIs, standardized data formats— Federated authentication and authorization, access control— Distributed storage — Mechanisms to search distributed information

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

21

Future Internet

> The current Internet is based on principles of the 1960s with the main purpose to interconnect large computers. Issues such as mobility, security, multimedia had lower importance.

> Today, Internet users are mainly interested in retrieving content and accessing services.

> Increasing bandwidth demand due to— (personalized) multimedia data streams— Cloud computing and storage,

visualization on (mobile / high-resolution) end systems— Huge amounts of data from measurements and scientific experiments— Internet of Things, Internet of Services, …

> New approaches, e.g., Information-Centric Networking, need more research, development, testing etc.

> World-wide Future Internet research programs, e.g., GENI, FIRE25. April 2012

Torsten Braun: Research Challenges in Networked Systems

22

Security

> Internet infrastructure security, e.g., DNS, routing> Security in new network paradigms, e.g., Future Internet> System security> Application security> Cloud computing security> Usability and security> Network forensics vs. privacy> Machine-to-machine communication,

smart grids, CPS> Federated social networks

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

23

Remarks on Research Organization

> Experimental research— Increasing importance of evaluation in testbeds,

e.g., sensor network conferences— shared research infrastructures and

repositories for results to– save costs– support comparability / repeatability of results

> Collaboration— High quality research benefits from joint research

activities with a critical mass of persons. – Example: ACM EuroSys 2012

– Average number of authors of accepted papers: 4.37– Average number of authors of rejected papers: 2.87

— But avoid massive coordination overhead and free riding> Evaluation

— Articles in CS conferences with low acceptance rates (15-30 %) have often more impact than most journals.

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

24

ICT in EU Horizon 2020

> New generation of components and systems including micro/nano-electronics and photonics technologies, components and embedded systems engineering

> Next generation computing, advanced computing systems and technologies

> Infrastructures, technologies and services for the future Internet

> Content technologies and information management including ICT for digital content and creativity.

> Advanced interfaces and robots and robotics and smart spaces

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

25

Conclusions

> Increasing importance of software, also in control and communication technology, possibly using cognitive approaches

> Security issues everywhere> Energy efficiency as a grand challenge

> Complexity requires (interdisciplinary) research collaboration

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

26

Statements about the Future II

> The best way to predict the future is to invent it.(Alan Kay)

25. April 2012

Torsten Braun: Research Challenges in Networked Systems

27

Thanks for your attention !

> Contact— [email protected] — http://cds.unibe.ch

25. April 2012