1 of 15

INODAM: RESILIENT ORCHESTRATION FOR AMAZONIAN ENVIRONMENTAL MONITORING

​

Graduation’s student

David Galhego

Advisor

Prof. Dr. Eduardo Cerqueira

2 of 15

The Amazon region faces extreme events like flash floods and riverbank erosion. Traditional physical sensors (limnímetros) are prone to failure due to obstruction by debris and mud. Relieable monitoring requires Computer Vision (CV) from edge cameras, but this introduces heavy data traffic in a resource-scarce environment.

2

3 of 15

INODAM

INODAM

3

4 of 15

Chain Flow

4

INODAM

5 of 15

5

6 of 15

6

7 of 15

7

8 of 15

8

9 of 15

Sensor Energy

(E

sen)

Since the Edge Server operates on unstable power sources, the system consolidates analysis requests. This maximizes throughput per watt, preventing the village's mini data center from energy collapse during critical events.

The cost function balances Sentinel longevity with Edge stability.

sen

9

sen

sen

10 of 15

Video Analysis Rate (%)

10

INODAM

INODAM

11 of 15

11

riverbank erosion

INODAM

12 of 15

12

INODAM

13 of 15

13

INODAM

INODAM

14 of 15

14

15 of 15

OBRIGADO

THANK YOU

15