engineering computational ecosystems (2nd year phd seminar)

2nd year Ph.D. Seminar

Engineering Computational Ecosystems

Danilo Pianini – [email protected]

Supervisor: prof. Mirko Viroli

Alma Mater Studiorum—Universita di Bologna

June 7, 2013

Danilo Pianini (UniBo) Computational Ecosystems June 7, 2013 1 / 78

Outline Every living creature on this earth dies alone.

1 Pervasive Computing(Near) Future city scenario(Near) Future devicesPervasive displaysCase studies

2 Pervasive EcosystemsCloud or pervasive?Self-organising spatial data structures

3 Engineering Pervasive EcosystemsAnticipative adaptationThe SAPERE EU ProjectAlchemistSimulations and demos

4 Open issuesProgramming the emergenceReal-world setup


Pervasive Computing

Outline






Pervasive Computing (Near) Future city scenario

Outline







Shibuya Crossing



People



(Near) Future city scenario


Pervasive Computing (Near) Future devices

Outline







Ready-to-use devices

Images courtesy of Alois Ferscha (Pervasive Computing Group, Johannes Kepler Universitat Linz)



Almost ready devices

Images courtesy of Alois Ferscha (Pervasive Computing Group, Johannes Kepler Universitat Linz)



Maybe in few decades...


Pervasive Computing Pervasive displays

Outline







More and more pervasive inside

Image courtesy of Alois Ferscha (Pervasive Computing Group, Johannes Kepler Universitat Linz)

→



More and more pervasive outside


. . . . . . .


Pervasive Computing Case studies

Outline







Pervasive displays: multiple view




Pervasive displays: local sharing




Pervasive displays: steering




Pervasive displays: crowd steering




Some interesting services and features

Display must adapt to people behaviour

Displays show information based on majority of people around

Alerts are shown as a given person passes nearby

Content redirected from smartphone to display

Adjacent displays show a common, bigger content

Displays coordinate to avoid irritating users

Displays coordinate to provide visualisation streams

Public displays sharedly used by friends to interact

Using eye-glasses for immersed interaction

People may change their behaviour because of that

Single User Steering via Public Displays

Crowd Balancing

Evacuation

Adaptive Advertisement

Collective Guidance


Pervasive Ecosystems

Outline






Pervasive Ecosystems Cloud or pervasive?

Outline







A standard, centralised SOA solution – cloud oriented



A standard, centralised SOA solution

A centralised solution

One service (typically in the cloud) for:

Discovery: what components are available in the system?

Context: where are components? (behaviour specialisation)

Orchestration: coordinating components

Data storage: depositing/retrieving data

Adaptation: reacting to contingencies

All components interact through such middleware services

Does it scale well?

Can displays promptly react?

Can we store streams of sensor data in the cloud?

What about privacy?



Eco-inspired SOA solution



Eco-inspired SOA solution

Fully decentralised middleware services

Locations become very small and form a huge dynamic set

Contextualisation, discovery, and orchestration almost vanish

Middleware service just as a single space distributed on all nodes

In overall we have a network of spaces with service “tags”

Adaptation is achieved by simple rules combining tags

Drawing a bridge with natural ecosystems

We have a set of spatially situated entities interacting according towell-defined set of natural laws enforced by the spatial environment inwhich they situate, and adaptively self-organizing their interactiondynamics according to their shape and structure[Zambonelli and Viroli, 2011]



Towards a computational continuum

A pervasive (eco)system:

is composed of myriads of devices (nodes)

contains devices located in space and time

has a notion of “node density”

As such, it is similar to particle systems, and the higher the density, themore this system approximates a spatio-temporal computationalcontinuum.


Pervasive Ecosystems Self-organising spatial data structures

Outline







Computational fields

A way to engineer self-organising applications in different scenarios

Sensor networks (aggregating/diffusing data)[Beal and Bachrach, 2006]

Swarm and modular robotics (cooperative team work)[Bachrach et al., 2010]

Pervasive computing (crowd steering, coordinated visualisation)[Viroli et al., 2011]

Rough idea

Two simple rules for creating a spatial structure:

1 Diffuse spatial enriched information among neighbours

2 Aggregate the information received by neighbours considering thespatial information



A 3D rendering of a uniform gradient structure



Towards a library of gradient-based computational fields

Gradient (discrete, flat continuous, structured continuous)

Other structures (channel, shrinking crown, sector, partition, shadow, . . . )



Creating optimal (multi-)paths

Enable devices used to steering from a source to a target

Open issue: fields react to current event, but they do not provideanticipative adaptation.

If obstacles are dynamically added and moved

If there is knowledge about WHEN those events will happen

How to improve?


Engineering Pervasive Ecosystems

Outline






Engineering Pervasive Ecosystems Anticipative adaptation

Outline







When spatial gradient is not enough



Anticipative gradient [Montagna et al., 2012a]

Anticipative gradient merges three strategies to provide anticipativeadaptation: in case a future event is detected

In case a future event is on the optimal path, also store paths notpassing through it

If you are in an area in which, if you followed the gradient, you wouldhave to wait

Either circumvent (green) or proceed and wait (red)


Engineering Pervasive Ecosystems The SAPERE EU Project

Outline







The SAPERE Project

http://www.sapere-project.eu


http://www.sapere-project.eu


Eco-laws and Live Semantic Annotations

Live Semantic Annotations (LSA)

A unified description for devices, data, services

Is about interface, status, and behaviour of a component

It provides semantic information, and it is dynamic

Eco-Laws

They resemble chemical reactions

They take some reagent LSA, and provide some product LSA

They can diffuse an LSA in the neighborhood

They can aggregate LSAs like in chemical bonding

They form a small & fixed set of natural eco-laws



Pervasive Ecosystems

LSA Space

LSA

LSALSA

LSA

eco-lawEngine

Software services

SAPERE Node

eco-law activation

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode

SAPERENode


Engineering Pervasive Ecosystems Alchemist

Outline







Computational ecosystems: functional guarantees

Houston, we’ve got a problem.

Prediction difficulties

Thousands-devices scale system

Autonomous devices

Interaction plays a huge role

Formal proof

Mathematical guarantee

Only tackles simple cases

Model checking

Space state explosion for non-trivial cases



Computational ecosystems: simulation

Open issue! How to simulate a (bio)chemical inspired computationalecosystem?

Classic ABM modelling

High flexibility

Topology as first-class abstraction

Dynamics explicitly modelled

No native support for CTMC model

Chemical inspired modelling

Natively CTMC

Very fast and reliable algorithms exist in literature

Very limited topology: multicompartment at best

Only classic reactions can change the world status: limited flexibility



Alchemist simulation approach [Pianini et al., 2013]

Base idea

Start from the existing work with stochastic chemical systemssimulation

Extend it as needed to reach desired flexibility

Goals

Full support for Continuous Time Markov Chains (CTMC)

Support for differently distributed events (e.g. Triggers)

Rich environments, with obstacles, mobile nodes, etc.

More expressive reactions

Fast and flexible SSA engine



Enriching the environment description

Environment

Node

Reactions

Molecules

Alchemist world

The Environment contains and links together Nodes

Each Node is programmed with a set of Reactions

Nodes contain Molecules

Each Molecule in a node is described with a ConcentrationDanilo Pianini (UniBo) Computational Ecosystems June 7, 2013 50 / 78


Extending the concept of reaction

From a set of reactants that combine themselves in a set of products to:

Number of

neighbors<3

Node

contains

something

Any other

condition

about this

environment

Rate equation: how conditions

influence the execution speed

Conditions Probability distribution Actions

Any other

action

on this

environment

Move a node

towards...

Change

concentration

of something

Reaction

In Alchemist, every event is an occurrence of a Reaction



SSA Algorithms

Several SSA exist, they follow the same base schema [Gillespie, 1977]:

1 Select next reaction using markovian rates

2 Execute it

3 Update the rates which may have changed



Existing algorithms

Next Reaction [Gibson and Bruck, 2000]

1 Compute a putative execution time for each reaction

2 Store reactions in a binary heap

3 Pick the next reaction

4 Execute it

5 Compute putative times only for potentially involved reactions

6 Goto 3



Smart data structures ⇒ bleeding edge performances

2.04 4

3.71

7.32 1

5.51 0

2

8.91 0

4.20 0

9.10 0

10.10 0

inf0 0

A+B→C

B+C→D E+G→A

D+E→E+F F→D+G

Next Reaction efficient structures made dynamic

Dynamic Indexed Priority Queue

Allow to access the next reaction to execute in O(1) timeWorst case update in log2 (N) (average case a lot better)Extended to ensure balancing with insertion and removal

Dynamic Dependency Graph

Allows to smartly update only a subset of all the reactionExtended with the concept of input and output context



Modular structure

Environment

User Interface

Alchemist language

Application-specific Alchemist Bytecode Compiler

Environment description in application-specific language

Incarnation-specific language

Reporting System

Interactive UI

Reaction Manager

Dependency Graph

Core Engine

Simulation Flow Language Parser

Environment Instantiator

XML Bytecode



Computational ecosystems: Alchemist



Some Features in short

Kinetic Monte Carlo engine

Parallelised engine

Mobility support

Dynamic connectivity support

Complex data items

Reaction-like programming

Parallel executor

Built-in approximate stochastic model checker

Alchemist2Blender

PVeStA integration

Built from scratch



Testbed

We realised the same crowd-steering application for both Alchemistand RePast, in order to evaluate the performance gap (if any)

Source code for RePast and standalone Alchemist applicationavailable athttp://www.apice.unibo.it/xwiki/bin/view/Alchemist/JOS

We choose a simplified use case which allows simulation in RePastwithout any change in its engine

This actually cripples out the most important Alchemist optimizationfor complex environment (the dependency graph)


http://www.apice.unibo.it/xwiki/bin/view/Alchemist/JOS


Results

0

50

100

150

200

250

300

350

400

450

500

50 100 150 200 250 300 350 400 450 500

Exe

cuti

on t

ime [

s]

Number of agents

Performance comparison with Repast

Repast Alchemist


Engineering Pervasive Ecosystems Simulations and demos

Outline







Crowd evacuation



Crowd steering



Crowd steering with anticipative adaptation



Realistic pedestrians



Morphogenesis



Morphogenesis [Montagna et al., 2012b]


Open issues

Outline






Open issues Programming the emergence

Outline







Abstraction level

We can not (yet) program the emergence.

Chemical-inspired rules are powerful, but hard to control

Interactions among zilions of possibly mobile nodes may exposeunpredicted situations

We would like to express a spatial pattern, and automatically get thesingle node program



Simulator improvement

Functional guarantees prior-to-deployment: large scale realistic simulations

Real life data (e.g. from Wien marathon)

Realistic environment (e.g. from Google Maps)

Realistic human behaviour (already partially there)


Open issues Real-world setup

Outline







Communication

P2P Communication issues

Bluetooth: tight bandwidth, slow device detection

ZigBee: sensor oriented, not enough diffused

WiFi: difference between master mode and “normal” mode

Near Field Communication: too Near!

Wifi direct: we need to deepen

What do you suggest?



Localisation and orientation

Outdoor

GPS: very good

Computer vision: good if the seeing node knows where it is

Latitude API: Android only, mixes 3G, WiFi, GPS and onboardmagnetometer. Very good

Indoor

GPS: useless within buildings

Computer vision: good if the seeing node knows where it is

Latitude API: Inaccurate, but promising

Signal power (Bluetooth, WiFi): Almost useless

Ultra Wide Band: extremely promising, but requires specific setup

Resonant coupling [Pirkl and Lukowicz, 2012]: promising!

Ideas?Danilo Pianini (UniBo) Computational Ecosystems June 7, 2013 73 / 78


Other issues

Identification

Relation between devices and userDevice identification: addresses, idsUser identification: computer vision, explicit login

Privacy

Security



Bibliography I

Bachrach, J., Beal, J., and McLurkin, J. (2010).Composable continuous-space programs for robotic swarms.Neural Computing and Applications, 19(6).

Beal, J. and Bachrach, J. (2006).Infrastructure for engineered emergence on sensor/actuator networks.IEEE Intelligent Systems, 21(2):10–19.

Gibson, M. A. and Bruck, J. (2000).Efficient Exact Stochastic Simulation of Chemical Systems with ManySpecies and Many Channels.The Journal of Physical Chemistry A, 104(9):1876–1889.

Gillespie, D. T. (1977).Exact stochastic simulation of coupled chemical reactions.Journal of Physical Chemistry, 81(25):2340–2361.



Bibliography II

Montagna, S., Pianini, D., and Viroli, M. (2012a).Gradient-based self-organisation patterns of anticipative adaptation.In Proceedings of 6th IEEE International Conference on Self-Adaptiveand Self-Organizing Systems (SASO 2012), pages 169–174.

Montagna, S., Pianini, D., and Viroli, M. (2012b).A model for drosophila melanogaster development from a single cell tostripe pattern formation.In Shin, D., Hung, C.-C., and Hong, J., editors, Proceedings of the27th Annual ACM Symposium on Applied Computing (SAC 2012),pages 1406–1412, Riva del Garda (Trento), Italy. ACM.

Pianini, D., Montagna, S., and Viroli, M. (2013).Chemical-oriented simulation of computational systems withAlchemist.Journal of Simulation.



Bibliography III

Pirkl, G. and Lukowicz, P. (2012).Robust, low cost indoor positioning using magnetic resonant coupling.In UbiComp, pages 431–440.

Viroli, M., Casadei, M., Montagna, S., and Zambonelli, F. (2011).Spatial coordination of pervasive services through chemical-inspiredtuple spaces.ACM Transactions on Autonomous and Adaptive Systems, 6(2):14:1 –14:24.

Zambonelli, F. and Viroli, M. (2011).A survey on nature-inspired metaphors for pervasive serviceecosystems.International Journal of Pervasive Computing and Communications,7(3):186–204.


SAPERE Self-aware Pervasive Service Ecosystems

Engineering Computational Ecosystems

Danilo Pianini – [email protected]

Supervisor: prof. Mirko Viroli

Alma Mater Studiorum—Universita di Bologna

June 7, 2013


engineering computational ecosystems (2nd year phd seminar)

Technology