Presentazione standard di PowerPointOperating Platforms Southbound interfacing Vision and Model Trends Examples of activities Service unification paradigm: i .e , SDN controller, network

GRUPPO TELECOM ITALIA

Future Internet:

towards “organisms” dynamics

IIT DAY

Pisa, 17 Novembre 2016

Antonio Manzalini, Ph.D

Technological Scouting, Trend Analysis & Future Center

2IIT DAY – Future Internet

Antonio Manzalini, TIMSource: https://www.wired.com/2015/06/mapping-

the-internet/


Antonio Manzalini, TIM

Source: https://www.linkedin.com/pulse/20140705094524-10816248-forget-cloud-computing-fog-computing

Internet: an enormous number of interconnected “things”



Source: http://digitalsynopsis.com/buzz/internet-real-time-infographic/

Internet: the creation and transfer of a huge amount of data



…but also: local “cognition” at the edge (i.e., future terminals)

http://www.google.it/url?sa=i&rct=j&q=&esrc=s&frm=1&source=images&cd=&cad=rja&uact=8&docid=SfxhxCRr9YH40M&tbnid=_ivKwh06vp2WTM:&ved=0CAUQjRw&url=http://robotnieuws.blogspot.com/&ei=3pVjU9_QI8SZPeDYgbAP&bvm=bv.65788261,d.bGE&psig=AFQjCNEB_bJGt_XSigYCLdpIBwPEM6Oj5A&ust=1399121543062157


6

LOGO PARTNER

…and a global interconnected “cognition” (i.e., SDN and NFV)

Environment

Senses

Organs

Effectors

Organs

percepts actions

Global cognition

(Cloud Edge DC)

IIT DAY – Future Internet


Local cognition

(terminals, Fog devices)

Programmable NetworksProgrammable Networks



7

LOGO PARTNER

• …with a sort of “nervous

system”:

• sensing, collecting and storing

(even locally, in the Edge-Fog

nodes) massive data sets (through

terminals, smart things,

intelligent machines);

• Moving sets of data or code

(through high bandwidth and ultra-

low low latency connections) where

it is more convenient (execution

of virtual functions and services

);

• elaborating said big data (Cloud

and Edge and Fog Computing) to

infer decisions for actuating

local actions (pervasive machines,

IoT, etc).

Then Internet is becoming like an “organism” …



Picture Credits:

http://www.eucognition.org

• It brings Adaptability at the cost of

Complexity.

8

An “organism” … providing any sort of “services”

• “Adaptability” is achieved by solving optimization

problems, e.g., constraint-based optimal allocation of

logical resources and links.

• Services can be seen as “graph” of components (e.g.,

software tasks) allocated/executed/moved in one or

more “network slice” (e.g., VMs or Containers,

interconnected via logical links);



9

…towards X-as-a-Service

• “X-as-a-Service” means that SDN Controllers,

Virtual Network Functions (VNF) and end-Users’

services/applications can be all seen as “apps”

under some generalized assumptions.

• Towards an unifying model of “X-as-a-Service”:

• provides a function (both “global” and “local”);

• exports APIs (e.g., REST);

• available anywhere and anytime (location-time

independent);

• scalable, elastic, and resilient;

• runs in a slice;

• can be built by composing with existing services;

• building blocks (NoSQL DB), user-facing (Facebook), and

both (DropBox).

• Need of looking at Future Internet with a

“systemic” approach





What does “X-as-a-Service” allow?

• Integration of SDN-NFV, Cloud-Edge Computing, Artificial

Intelligence and Big Data will allow providing any Services for

automating/optimising Digital Society and Economy.Source: http://www.androidauthority.com/arm-internet-of-things-385616/,

modified

Cloud -

Edge -

Computing

NFVSDN

A.I.

Big

Data

http://www.androidauthority.com/arm-internet-of-things-385616/

11

“Adaptability” comes at a cost: more “Complexity”

• We’re evolving from managing sets of “closed”

physical nodes and systems to orchestrating a huge

number of software entities, logically intertwined

and dynamically moving.

• This "complexity", outstripping human control

ability and configuration, and may cause “non

linearity” and instabilities.Complexity is a property

of Future Internet

Lower Bound of Complexity

(predictable but static, brittle)

Upper Bound of Complexity

(dynamic, more dependable but unpredictable)

How creating the right “constraints” ?



12

Operating System to create “Constraints that Deconstrain…”

Source: J. Doyle «Universal Laws and Architectures»



“… it’s a Universal Law…”

13

…sort of Operating System also in “organisms” (e.g., bacteria)



14

Quest for cognitive Operating Systems/Platforms

Northbound

interfacing

Applications/Services(X-as-a-Service)

Infrastructure(including IT, Network and Terminals)

Real Time Cognitive

Operating Platforms

Southbound

interfacing

Vision and Model Trends Examples of activities

Service unification

paradigm: i.e., SDN

controller, network

function, cloud services

are just like “apps” to be

execute on an Operating

System

Automated of Operation

processes

Agile end-to-end

provisioning

Security by design

OpenStack, ONF/ONOS,

CORD, ETSI NFV,

OpenMANO, MesoSphere,

etc

Cloudlet, OpenFog, ETSI

MEC, IEEE SDN “OMEC”

etc

Commoditization (e.g.,

Central Offices to become

DC based on almost standard

h/w);

Disaggregation Into

components

Open Compute Project

(OCP) Telecom Infra

Project (TIP)

IEEE SDN “OMEC”

ETSI MEC, etc

Intent framework (IETF)

TOSCA/YANG (OASIS,

IETF)

OpenFlow, NetConf, …,

P4, etc.

Simplification of

interfacing and

unification of templates

Programming languages and

configuration protocols

X-OS/CORD

5GPPP



15

… some clear evidences



16

… some clear evidences

https://wtvox.com/robotics/google-is-working-on-a-new-algorithm-thought-vectors/?utm_source=dlvr.it&utm_medium=twitter

• Prof. Geoff Hinton working at Google to develop a

cognitive OS, equipped with algorithms to encode

thoughts as sequences of numbers (“thought vectors”):

target is an OS with human-like logic and capacity for

reasoning.



https://wtvox.com/robotics/google-is-working-on-a-new-algorithm-thought-vectors/?utm_source=dlvr.it&utm_medium=twitter

17

… some clear evidences: towards an A.I. Internet…



Ambients that

“learn”• http://www.forbes.com/sit

es/freddiedawson/2016/05/

24/the-house-that-learns-

bringing-artificial-

intelligence-into-the-

home/#36f0e58483dc

Self-Driving Vehicles• https://www.google.com/selfdrivingc

ar/

Financial, Markets

predictions

Robots for HF trading• http://singularityhub.com/2016/06/01/a

re-high-stress-decisions-best-made-by-

bots/?utm_content=buffer4c953&utm_me

dium=social&utm_source=twitter.com&ut

m_campaign=buffer

Genetics, Designing

properties of drugs• http://www.eurekalert.org/pub_rel

eases/2016-03/imi-plt031516.php

Jeff Bezos (CEO of Amazon): “A.I. is

going to become gigantic and it's

probably difficult to overstate the

impact it will have on society over

the next 20 years“

• http://www.mediapost.com/publications/article/2770

96/jeff-bezos-says-ai-could-become-amazons-fourth-

pi.html

http://www.forbes.com/sites/freddiedawson/2016/05/24/the-house-that-learns-bringing-artificial-intelligence-into-the-home/#36f0e58483dc

https://www.google.com/selfdrivingcar/

http://singularityhub.com/2016/06/01/are-high-stress-decisions-best-made-by-bots/?utm_content=buffer4c953&utm_medium=social&utm_source=twitter.com&utm_campaign=buffer

http://www.eurekalert.org/pub_releases/2016-03/imi-plt031516.php

http://www.mediapost.com/publications/article/277096/jeff-bezos-says-ai-could-become-amazons-fourth-pi.html

18

… it’s a matter of a universal set of abstractions



• Data flow graphs describe computation

with a directed graph of nodes &

edges.

• Nodes typically implement mathematical

operations, but can also represent

endpoints, to push out results, or to

read/write persistent variables.

• Edges describe the input/output

relationships between nodes, carrying

dynamically-sized multidimensional

data arrays, or tensors.

• Nodes are assigned to computing nodes

and execute asynchronously and in

parallel once all the tensors on their

incoming edges becomes available.https://www.tensorflow.org/

19

A glimpse into the future: key areas



• Definition of a universal set of Abstractions

• Services to be modeled as “complex graph of

operations”…dynamically allocated on a “continuum” of

logical resources;

• Machine-Deep Learning and A.I. in general to make use /

exploit this widely.

• Adoption techno-economic “systemic” perspectives

• Humans will be more and more “in the loop” steering Future

Internet dynamics, but also things, machines and various

“software” entities (e.g., A.I. Avatars) …

• “Layered and distributed” Operating Systems/Platforms (with

local-global cognition capabilities ) to set the equilibria

Adaptability vs Complexity.

• Design of Security: “immune system” is a good

metaphor !

• looking for data patterns, just as our immune system looks

for patterns to identify pathogens, by innate and adaptive

approaches.



https://www.ted.com/talks/tal_danino_we_can_use_bacteria_to_detect_cancer_and_maybe_treat_it/tr

anscript?language=en

“… my hope is that the beauty and purpose of this

microscopic universe can inspire new and creative

tech. approaches for the future…”

Learning from the Internet of … bacteria

“… there are more bacteria in your body than stars in

our entire galaxy !”

21

Grazie !

http://ieee-sdn.blogspot.it/

https://twitter.com/a_manzalini



http://ieee-sdn.blogspot.it/

https://twitter.com/a_manzalini

Presentazione standard di PowerPointOperating Platforms Southbound interfacing Vision and Model Trends Examples of activities Service unification paradigm: i .e , SDN controller, network

Documents