• Non ci sono risultati.

Performance analysis of an underwater acoustic network

N/A
N/A
Protected

Academic year: 2021

Condividi "Performance analysis of an underwater acoustic network"

Copied!
8
0
0

Testo completo

(1)

UNIVERSIT `

A DI PISA

Facolt`a di Ingegneria

Laurea Specialistica in Ingegneria dell’Automazione

Tesi di laurea

Performance analysis of an underwater acoustic network

Candidato:

Lorenzo Fusini

Relatore:

Prof. Andrea Caiti

Sessione di Laurea del 11/05/2012 Anno Accademico 2011/2012

(2)

Abstract

This thesis aims at analysing some of the many data collected during the final sea trial of the UAN11 project (Underwater Acoustic Network) in the Trond-heim fjord, Norway. The objective is to check how conditions of the underwa-ter acoustic channel affected communications, namely ping sequences, between specific nodes and, in those periods of time when good conditions were associ-ated with bad communication results, try to find the reason why. The available ping results were fetched and used to identify periods of time with good or bad communications between the master node and other fixed or mobile nodes: such results were compared with the channel impulse responses stored, record by record, in the acoustic modems logfiles. Whenever ping results and chan-nel conditions did not match, the upper communications levels (UANtun and MOOS) were analysed to try and find at which point the system failed. The matching was generally positive, but sometimes a favourable acoustic channel delivered bad ping results: a deeper analysis showed that apparently a traffic overload of the UANtun level was the cause for lost pings.

(3)

Table of Contents

1 Introduction 6

2 Background 8

2.1 The UAN project . . . 8

2.2 The sea trial in Trondheim fjord . . . 11

2.3 Overview on the available data from the sea trial . . . 13

3 Data reading and processing methods 14 3.1 How to read the data . . . 14

3.1.1 KM modems logfiles . . . 14

3.1.2 Ping results . . . 16

3.1.3 UANtun logfiles . . . 17

3.1.4 MOOS database . . . 18

3.2 Data presentation . . . 19

3.3 The CFAR algorithm . . . 20

3.4 SNIR definition . . . 21 4 Results 23 4.1 23 May . . . 23 4.1.1 FN2 (pier) . . . 24 4.1.2 FN1 (pier) . . . 25 4.2 24 May . . . 27 4.2.1 FN1 . . . 27 4.3 25 May . . . 29 4.3.1 FN2 . . . 29 4.4 26 May . . . 32 4.4.1 FN2 . . . 33 4.4.2 MN3 . . . 37 4.5 27 May . . . 39 4.5.1 FN2 . . . 39 4.5.2 MN1 . . . 41 4.5.3 MN3 . . . 43 2

(4)

4.6 UANtun and MOOS logfiles compared . . . 45

4.6.1 Select outgoing messages . . . 46

4.6.2 To Matlab . . . 47

4.6.3 Correspondences . . . 47

5 Discussions 50

6 Conclusions 52

A Received records sorted by day and source node 53

B An excerpt from a UANtun log 55

C An excerpt from the MOOS log 56

(5)

List of Figures

1 The test area in the Trondheim fjord. . . 12

2 RV Gunnerus. . . 12

3 Example of CIR, from record From10038(3729). . . 16

4 Example of ping sequence as presented by SINTEF. . . 17

5 Example of plots for inspecting channel conditions . . . 20

6 Deployment geometry for 23 May. . . 23

7 Contour map (top) and peak/SNIR values (bottom) for FN2, 23 May 17:37-20:15. . . 24

8 Ping sequence against FN2 on 23 May 17:35-20:15. . . 24

9 Ping sequence against FN2 on 23 May 20:15-22:00. . . 25

10 Contour map (top) and peak/SNIR values (bottom) for FN1, 23 May 16:14-22:55. . . 26

11 Ping sequence against FN1 on 23 May 20:13-22:35. . . 26

12 Deployment geometry for 24 May. . . 27

13 Contour map (top) and peak/SNIR values (bottom) for FN1, 24 May 14:06-18:58. . . 28

14 Ping sequence against FN1 on 24 May 14:51-15:00. . . 28

15 Deployment geometry for 25 May. . . 29

16 Contour map (top) and peak/SNIR values (bottom) for FN2, 25 May all day. 30 17 Ping sequence against FN2 on 25 May 10:26-10:38. . . 30

18 Ping sequence against FN2 on 25 May 11:06-21:50. . . 31

19 Records corresponding to the second ping sequence against FN2 on 25 May. 31 20 Deployment geometry for 26 May. . . 32

21 Contour map (top) and peak/SNIR values (bottom) for FN2, 26 May 22:00(25 May)-15:52. . . 33

22 Magnified version of Fig. 21. . . 34

23 Ping sequence against FN2 on 26 May 7:58-12:50. . . 34

24 Ping sequence against FN2 on 26 May 12:50-14:25. . . 35

25 Ping sequence against FN2 on 26 May 15:06-15:40. . . 35

26 Contour map (top) and peak/SNIR values (bottom) for FN3, 26 May 11:22-15:43. . . 36

(6)

27 Contour map (top) and peak/SNIR values (bottom) for MN1, 26 May 9:17-11:15. . . 37 28 Contour map (top) and peak/SNIR values (bottom) for MN3, 26 May

12:17-14:08. . . 38 29 Ping sequence against MN3 on 26 May 12:50-14:25. . . 38 30 Deployment geometry for 27 May. . . 39 31 Contour map (top) and peak/SNIR values (bottom) for FN2, 27 May

10:49-24:00. . . 40 32 Ping sequence against FN2 on 27 May 12:16-14:40. . . 40 33 Ping sequence against FN2 on 27 May 14:48-15:50. . . 41 34 Contour map (top) and peak/SNIR values (bottom) for MN1, 27 May

8:28-12:09. . . 42 35 Ping sequence against MN1 on 27 May 8:49-12:15. . . 42 36 Ping sequence against MN1 on 27 May 14:53-15:55. . . 43 37 Contour map (top) and peak/SNIR values (bottom) for MN3, 27 May

11:36-16:26. . . 43 38 Ping sequence against MN3 on 27 May, 12:15-14:45. . . 44 39 Ping sequence against MN3 on 27 May, 14:58-15:45. . . 44 40 MOOS/UANtun correspondences for STU-FN2 messages on 25 May,

8:04-24:00 UTC. . . 48 41 MOOS/UANtun correspondences for STU-FN2 messages on 26 May,

00:00-16:31 UTC. . . 48 42 MOOS/UANtun correspondences for STU-FN2 messages on 25 May,

10:25-21:50 UTC. . . 48 43 MOOS/UANtun correspondences for STU-FN2 messages on 26 May,

(7)

1

Introduction

Wireless underwater acoustic networks became possible in the late ’80s, following the application of newly discovered digital technologies to acoustic modems [?]. Until then underwater acoustic waves could not be used to efficiently convey information but only for detection purposes, as in sonars. The development of an underwater network of modem-equipped devices capable of exchanging data started in the early ’90s [?] and still constitues one of the most challenging subjects of marine acoustics.

The present thesis proposes to analyse the performance of one of such networks, namely the one developed during the UAN (Underwater Acoustic Network, often referred to as UAN11) project in 2010-11 by a group of companies and institutions from different Eu-ropean countries and partially funded by the EuEu-ropean Union under the 7th Framework Programme. The whole designed system, its components and functionalities were all tested during the final sea trial of the project, which took place in May 2011 in Trondheim fjord, Norway. A fjord represents a harsh environment for acoustic propagation: shallow wa-ter, water inflow from rivers, quickly varying bathymetry make the development of such a system very complex. An analysis of the measurements from the sea trial is of the ut-most importance for solving the problems connected to this particular kind of underwater network.

The work presented here was carried out in Trondheim in collaboration with research scientists Knut H. Grythe, Arne Lie and Tor Arne Reinen from SINTEF (Foundation for Industrial and Technical Research) and professor emeritus Jens M. Hovem from NTNU (Norwegian University of Science and Technology) in fall 2011.

Some of the many data collected during the final trial were analysed, mostly by means of Matlab. More specifically, at first the available results from the ping tests performed during the trial were compared with the acoustic channel conditions present at the same periods of time. Such data could be found in the modems logfiles: each record in the log reports several types of information, amongst which the channel impulse response measured at the moment. Unfortunately not all the logfiles were available: this, together with errors in the transmission records, hindered part of the work and made it impossible to draw conclusions for some of the underwater nodes.

After having selected some periods of time, the corresponding records were fetched and analysed. Being the logfiles very large this procedure was prone to mistakes, hence

(8)

time-consuming. Channel conditions and ping results usually matched, but sometimes a favourable channel delivered negative ping results. These special cases were further analysed at upper levels of communications (UANtun and MOOS) to see at which point the system failed. Again only a subset of all the necessary data was available, so conclusions are available only for one of the nodes.

Chapter ?? gives a background of the UAN11 project, its aim, main components and tested functionalities, a brief description of the final sea trial and an overview of the avail-able data. Such data are stored in the central server of the project, located in Portugal. Detailed reports of the whole project were available, though as yet restricted to the par-ticipating members, for gathering all the information necessary to carry out the present work.

Chapter ?? describes in detail the types of data used for the intended purposes of this thesis, how they could be read and processed in order to be easily analysed and how they are presented.

Chapter ?? reports the actual analyses, the comparisons between channel conditions and ping results and, when necessary and possible, a further investigation by means of the UANtun and MOOS logfiles. All this is done to a large extent using Matlab.

In Chapter ?? results are discussed, together with considerations on what more could be done before and during the sea trial for carrying out more accurate and conclusive analyses.

Riferimenti

Documenti correlati

Nel presente lavoro sono stati affrontati la preparazione e lo studio di un nuovo laminato costituito da polipropilene (PP) nanocomposito (NCP) rinforzato da una stuoia in fibra

In order to obtain a quantitative estimation of the advantages of the proposed continuous strategy, with respect to the batch one, a functional version of the continuous plant has

- Flash-GC was applied to discriminate VOOs with different geographical origin - Chemometric data elaborations with untargeted approach were realized.. - PLS-DA and

[r]

• Grazie a Wladimir Palant, AdBlock ha fatto un enorme passo avanti, diventando AdBlock Plus.. •AdBlock infatti non riusciva completamente a riconoscere e, di conseguenza, a

These samples are then used to compute the statistical parameters of each re- sponse value (i.e. mean value, standard deviation, skewness, kurtosis) and to evaluate the

• On average across OECD countries and in almost all countries/economies that participated in PISA 2018, students in disadvantaged schools had less access than students in

Keywords: chitosan; physico-chemical properties; molecular weight; degree of acetylation; conformation; antimicrobial activity; from gel to nanobeads; drug