Autonomous Mouvments: to creativity... and beyond Tahiry - - PowerPoint PPT Presentation

Tahiry Razafindralambo May 27, 2014

Autonomous Mouvments: to creativity... and beyond

I am a Inria researcher in the FUN research group led by our beloved Nathalie. I was ”out

• f the office” for the last 20 months.

Now, I am back. I (used to) work on: #Sensor Net- works, #Controlled Mobility, #QoS, #Algo- rithm, #Mobility. Please come to visit me in the beautiful office #A205 with a perfect view

• n #NothingForNow and sometimes #Traffic-
• Jam. (It’s a 8 years picture I have much less hair now).

Tahiry Razafindralambo May 27, 2014 - 2

Tahiry Razafindralambo

Agenda

Chile Context Research statement Mobility Coverage Points of Interest Coverage, Permanent Connectivity Perspectives

Tahiry Razafindralambo May 27, 2014 - 3

Chile

Chile

Tahiry Razafindralambo May 27, 2014 - 4

Chile

What I did in Chile :: 1

I was part of Inria Chile

◮ It’s like Inria but in Chile ◮ It focuses only on technology transfer ◮ Only STI and SED ◮ Researchers dedicated to transfer / projects

Tahiry Razafindralambo May 27, 2014 - 5

Chile

What I did in Chile :: 2

Writing projects

◮ Health monitoring ◮ Enhanced shopping experience ◮ Precision agriculture ◮ Mines monitoring ◮ ...

Tahiry Razafindralambo May 27, 2014 - 6

Context

1

Autonomous movement

Tahiry Razafindralambo May 27, 2014 - 7

Context Research statement

Research Statements

◮ Protocols : MAC, Routing (wireless ad hoc networks, sensor networks) ◮ Performance evaluation : Process algebra / Model (ad hoc networks) ◮ Data aggregation : Temporal series (sensor networks) ◮ Controlled Mobility (sensor networks, wireless networks)

Tahiry Razafindralambo May 27, 2014 - 8

Context Context

Mobility::Definition

It is a characteristic of an object with the capacity or possibility to :

◮ Change aspect ◮ Move ◮ Be moved

Mobility characterizes everything that :

◮ Can “rapidly” change its shape ◮ Is instable ◮ Is variable ◮ Fluctuates over time and space

Tahiry Razafindralambo May 27, 2014 - 9

Context Context

Focus::Challenges

◮ Exploit the mobility of devices ◮ Introduce mobility inside Swarms/Networks ◮ Turn devices’ mobility into network/application primitive

Tahiry Razafindralambo May 27, 2014 - 10

Coverage

2

Coverage

Tahiry Razafindralambo May 27, 2014 - 11

Coverage PoI

Assumptions::Devices/Object

Devices / Robots

◮ Knows its own position ◮ Has limited communication capabilities (range R) ◮ Has sensing capabilities ◮ Has motion capabilities ◮ Decides to move on its own

Object / Network

◮ Knows the position of the Points of Interest ◮ Needs information from the Points of Interest ◮ Sends commands to the devices

Tahiry Razafindralambo May 27, 2014 - 12

Coverage PoI

Connectivity::Why ? How ?

Why

◮ Because it is possible ◮ Important to send/receive commands ◮ Important for data gathering ◮ Important for communication in general

How

◮ Based on local interactions ◮ Use of Relative Neighborhood Graph ◮ Add constraints on movements

We are using mobility to control and create the shape of the network. A short video...

Tahiry Razafindralambo May 27, 2014 - 13

Coverage PoI

Algorithm::Connectivity Part::1

• T. Razafindralambo, D. Simplot-Ryl: Connectivity Preservation and Coverage Schemes for Wireless Sensor Networks. IEEE Transaction on Automatic
• Control. 56(10): 2418-2428 (2011)

◮ We consider that the network is a graph (evolving) G(V , E) ◮ d(u, v) is the distance between devices u and v ◮ E = {(u, v) ∈ V 2 | u = v ∧ d(u, v) ≤ R} ◮ N(u) = {v ∈ E | d(u, v) ≤ R} neighbor’s of u ◮ RNG(G) = (V , E rng) RNG graph extracted from G. ◮ E rng = {(u, v) ∈ E | ∄w ∈ (N(u) ∩ N(v)) ∧ d(u, w) <

d(u, v) ∧ d(v, w) < d(u, v)}

◮ d+(u) distance between device u and its furthest RNG neighbor.

Tahiry Razafindralambo May 27, 2014 - 14

Coverage PoI

Algorithm::Connectivity Part::2

• T. Razafindralambo, D. Simplot-Ryl: Connectivity Preservation and Coverage Schemes for Wireless Sensor Networks. IEEE Transaction on Automatic
• Control. 56(10): 2418-2428 (2011)

Theorem

Let d ≤ (R − d+(u))/2 the maximum distance device u can travel. If at t1 G(V , E) is connected, at time t2 > t1 G(V , E) will be connected.

Intuition

u v w d (u) + R d d

Tahiry Razafindralambo May 27, 2014 - 15

Coverage PoI

Algorithm::PoI Coverage::0

• M. Erdelj, T. Razafindralambo, D. Simplot-Ryl: Covering Points of Interest with Mobile Sensors. IEEE Transaction Parallel Distributed Systems 24(1):

32-43 (2013)

Algorithm has 3 independent parts

◮ Direction (objectives) ◮ Connectivity preservation (constraints) ◮ Movement

Tahiry Razafindralambo May 27, 2014 - 16

Coverage PoI

Algorithm::PoI Coverage::1

• M. Erdelj, T. Razafindralambo, D. Simplot-Ryl: Covering Points of Interest with Mobile Sensors. IEEE Transaction Parallel Distributed Systems 24(1):

32-43 (2013)

Device movement sketch

◮ Move toward point of interest ◮ Consider connectivity preservation ◮ Stop when the PoI is reached (below sensing range) ◮ ... (special cases related techniques)

dp v w

R

u Base Station PoI x w PoI

R

u v Base Station x dp w PoI

R

u Base Station v x PoI

R

u w Base Station x v Tahiry Razafindralambo May 27, 2014 - 17

Coverage PoI

Algorithm::PoI Coverage::2

• M. Erdelj, T. Razafindralambo, D. Simplot-Ryl: Covering Points of Interest with Mobile Sensors. IEEE Transaction Parallel Distributed Systems 24(1):

32-43 (2013)

Properties

◮ Termination: There exists a time tend where all devices stop moving ◮ Straight line: The devices will form a straight line (at the end) ◮ Minimizing communication devices ◮ Maximizing sensing devices

Tahiry Razafindralambo May 27, 2014 - 18

Coverage PoI

Algorithm::PoI Coverage::3

• M. Erdelj, T. Razafindralambo, D. Simplot-Ryl: Covering Points of Interest with Mobile Sensors. IEEE Transaction Parallel Distributed Systems 24(1):

32-43 (2013)

2 4 6 8 10 12

• 1
• 0.5

0.5 1 1.5 2 2 4 6 8 10 12

• 1
• 0.5

0.5 1 1.5 2 2 4 6 8 10 12

• 1
• 0.5

0.5 1 1.5 2 2 4 6 8 10 12

• 1
• 0.5

0.5 1 1.5 2

Tahiry Razafindralambo May 27, 2014 - 19

Coverage PoI

Algorithm::Discussions

Strengths

◮ Simple, localized ◮ Separation between connectivity, coverage and movement ◮ Proven theoretically and practically ◮ Can be extended easily ◮ ...

Weaknesses

◮ Localization ◮ Need the location of Point of Interest ◮ ...

Tahiry Razafindralambo May 27, 2014 - 20

Perspectives

4

Perspectives

Tahiry Razafindralambo May 27, 2014 - 21

Perspectives

Next steps :: Controlled mobility::1

Why

◮ Decide if the movement is worth it ◮ Decide if the movement will have an impact ◮ Controlled Delay Tolerant Networks ◮ Relationship between movement/location and objectives

When

◮ Decide the best moment to move ◮ Collect the information that allows the decision ◮ Measurement system ◮ Forecasting techniques

Tahiry Razafindralambo May 27, 2014 - 22

Perspectives

Next steps :: Controlled mobility::2

Where

◮ Impact/Influence of position on the object ◮ Local choices and global needs ◮ Link the local position to a global optimum ◮ Define, change, update the objectives

How

◮ Choose the best way to move from one point to another ◮ Decide the interactions with other devices ◮ Swarm behaviour ◮ Trade-off : Energy-Objective

Tahiry Razafindralambo May 27, 2014 - 23

Perspectives

Next steps :: Controlled mobility::3

I have 5 minutes video...

Tahiry Razafindralambo May 27, 2014 - 24

Perspectives

MERCI

Tahiry Razafindralambo tahiry.razafindralambo@inria.fr http://researchers.lille.inria.fr/∼razafind/