Optimizing Network Performance through Packet Fragmentation in Multi- hop Underwater Communications Stefano Basagni ∗, Chiara Petrioli † Roberto Petroccia † Milica Stojanovic ∗ ∗ ECE Department Northeastern University †Dipartimento di Informatica Universit`a di Roma “La Sapienza” IEEE OCEANS 2010

Outline Introduction Data packet fragmentation Performance evaluation Conclusion 1

Introduction Underwater wireless networking has been recognized for a wide spectrum of applications  Scientific exploration  Commercial exploitation  Coastline protection  Prediction of underwater seismic  Prediction of volcanic events 2

Introduction Underwater wireless networking  Acoustic propagation  Access the channel using MAC protocol 3

Introduction Underwater acoustic networks  Long propagation delay  Low bandwidths 4 TIME Sender Receiver TerrestrialUnderwater Propagation delay

Introduction MAC protocol for underwater acoustic network  Distance-Aware Collision Avoidance Protocol(DACAP)  CSMA  MACA  RTS/CTS 5 B. Peleato and M. Stojanovic, “Distance aware collision avoidance protocol for ad-hoc underwater acoustic sensor networks.”IEEE Communications Letters, vol. 11, no. 12, pp. 1025–1027,December 2007.

A Introduction Collisions due to two nodes cannot hear each other directly  65% are control/data  10% are data/data  15% are among control packet 6 C B D collision

Motivation Data collision  Entire data may have to be discarded 7

Goal Partitioning a long packet into smaller fragments  Using selective repeat ARQ Reducing control/data collision Reducing over traffic 8

9 Data packet fragmentation

Assumption Each node has long data to send Each node fragments a data packet into k fragments  Each fragment with an Fragments are sent in a group and a cumulative ACK in return  i-bit of the ACK  1 for received correctly  0 for otherwise 10 ACK [1,1,1,0,1] Group

Data packet fragmenting Method 1 Method 2 11

Data packet fragmenting Method 1  Each transmitter sends out a group of k fragments and awaits the ACK  Fragments with negatively acknowledged are re- transmitted in the next group  No new packet fragment are sent along with those re-transmitted 12 TIME S D RTS CTS ACK [1,1,0,0,1] ACK [1,1,0,0,1] Interfering packet TIME S D RTS CTS ACK [1,1,] ACK [1,1,]

Data packet fragmenting Method 2  Rule 1  If the number of old fragments waiting to be sent is h, with h ≤ k/5,then a group of k + h fragments is sent  Rule 2  Old fragments are sent at the end of the group 13

Data packet fragmenting Method 2 14 S D RTS CTS ACK [1,1,0,1,1] ACK [1,1,0,1,1] Interfering packet TIME S D RTS CTS ACK [1,1,1,1,1,1] ACK [1,1,1,1,1,1] TIME 3 3 CTS

15 Performance Evaluation

SimulatorNS-2 Numbers of node at a depth of 200m100 static Distribution randomly & uniformly at depth of 200m Transmission radius1000m Number of fragments5 to 100, increment of 5 RouteShortest path Receiving power & idle power100mW Transmission power4930mW Transmission rate2000bps Data payload size3000Byte Traffic0.01 to

Performance metrics Throughput efficiency Energy per bit End-to-end latency per meter Route length 17

Performance metrics Throughput efficiency 18

Performance metrics Energy per bit 19

Performance metrics End-to-end latency per meter 20

Performance metrics Route length 21

Conclusion The use of packet fragmentation and selective repeat ARQ was considered for random access underwater networks  Throughput efficiency  Reducing end-to-end latency  Energy per bit consumption 22

Wireless Access and Networking Technology Lab (WANT Lab) Thanks for your attention 23 T h f r u o y a t e nt i o n t r o a n k s Wireless Access and Networking Technology Lab (WANT Lab)

Thanks for your attention 24 T h f r u o y a t e nt i o n t r o a n k s