Networks May 1, 2017 Imad Jawhar 1 , Sheng Zhang 2 , Jie Wu 3 , - PowerPoint PPT Presentation

Efficient Topology Discovery and Routing in Thick Wireless Linear Sensor Networks May 1, 2017 Imad Jawhar 1 , Sheng Zhang 2 , Jie Wu 3 , Nader Mohamed 4 , and Mohammad M. Masud 5 1 Midcomp Research Center, Saida, Lebanon 2 State Key Laboratory for Novel Software Technology, Nanjing University, P. R. China 3 Dept. of Comp. and Inf. Sciences, Temple University, Philadelphia, PA, USA 4 Middleware Technologies Labs, Isa Town, Bahrain 5 College of Information Technology, United Arab Emirates University, Al Ain, UAE

Outline l Introduction : Linear Sensor Networks ( LSNs ). Applications and architectures l Thick LSN model and definitions l Algorithms for backbone discovery in thick LSNs l Simulation and results l Conclusions and future research

Linear Sensor Networks (LSNs) Wireless sensor networks ( WSN ) l advancements in technology Sensor networks application : l environmental, military, agriculture, inventory control, healthcare, etc. Existing WSN research is 2-D or 3-D l deployment. Assumption that the network used for l sensors does not have a predetermined structure . Basic Senor Data Relay Data Node (BSN) Node (DRN) Dissemination Linear alignment of sensors can arise l Node (DDN) in many applications Linear characteristic can be utilized for l enhancing the routing and reliability in the such systems. We can design adapted protocols for l this special kind of sensor networks . Basic Sensor Basic Sensor Node (BSN)

Applications of LSNs Oil, Gas, and Water Pipeline Monitoring l Border Monitoring l IVC Network l Railroad/subway monitoring l Other applications: River, and sea cost l monitoring, etc.

Graph-Search-Based Topology Discovery Algorithm for LSNs Nodes identify nodes to be included in the backbone to reach the sink. l Backbone discovery increases the efficiency , and robustness of the network. l Allows more scalability of communication along LSN which can have large number of l nodes (hundreds or thousands) Can enhance reliability by “ jumping ” over failed by increasing communication range. l No need for location detection (e.g. GPS ), with higher cost and complexity of SNs . l Linear Backbone Discovery l (LBD) Algorithm F Node at primary edge sends G V l W Z X LD(4, ABCE) LD(5, ABCEG) LD(5, ABCHJ) Linear Discovery ( LD ) message . LD(4, ABCE) LD(7, ABCIKLU) S E LD(7, ABCIKLU) LD(4, ABCH) U LD(5, ABCHJ) T Message ID : to prevent looping LD(6, ABCIKL) l LD(2, AB) LD(7, ABCIKLM) LD(3, ABC) H J Sink LD(5, ABCHJ) Discovery Initiator Node LD(3, ABC) SF(7, ABCIKLM) myID : ID of sending node l C B LD(2, AB) A LD(1, A) L LD(5, ABCIK) LD(3, ABC) LD(6, ABCIKL) LD(7, ABCIKLM) K LD(4, ABCI) M LD(2, AB) I LD(5, ABCIK) messageLC : linear discovery l LD(4, ABCI) Q Y counter . Current count from LD(6, ABCIKY) LD(3, ABQ) LD(6, ABCIKY) N primary edge node. R LD(5, ABCIN) Sink Found (SF) Message O LD(6, ABCINO) PATH : ordered list of nodes P l Linear Discovery (LD) Message contained in discovered path

LD Message Propagation – LBD Algorithm F G V W Z X LD(5, ABCEG) LD(4, ABCE) LD(5, ABCHJ) LD(4, ABCE) LD(7, ABCIKLU) S E LD(7, ABCIKLU) LD(4, ABCH) U LD(5, ABCHJ) T LD(6, ABCIKL) LD(2, AB) LD(7, ABCIKLM) LD(3, ABC) H J Sink LD(5, ABCHJ) Discovery Initiator Node LD(3, ABC) SF(7, ABCIKLM) C B LD(2, AB) A LD(1, A) L LD(5, ABCIK) LD(3, ABC) LD(6, ABCIKL) LD(7, ABCIKLM) K LD(4, ABCI) M LD(2, AB) I LD(5, ABCIK) LD(4, ABCI) Q Y LD(6, ABCIKY) LD(3, ABQ) LD(6, ABCIKY) N R LD(5, ABCIN) Sink Found (SF) Message O LD(6, ABCINO) P Linear Discovery (LD) Message

LD Message Propagation – LBD Algorithm F G V W Z X LD(5, ABCEG) LD(4, ABCE) LD(5, ABCHJ) LD(4, ABCE) LD(7, ABCIKLU) S E LD(7, ABCIKLU) LD(4, ABCH) U LD(5, ABCHJ) T LD(6, ABCIKL) LD(2, AB) LD(7, ABCIKLM) LD(3, ABC) H J Sink LD(5, ABCHJ) Discovery Initiator Node LD(3, ABC) SF(7, ABCIKLM) C B LD(2, AB) A LD(1, A) L LD(5, ABCIK) LD(3, ABC) LD(6, ABCIKL) LD(7, ABCIKLM) K LD(4, ABCI) M LD(2, AB) I LD(5, ABCIK) LD(4, ABCI) Q Y LD(6, ABCIKY) LD(3, ABQ) LD(6, ABCIKY) N R LD(5, ABCIN) Sink Found (SF) Message O LD(6, ABCINO) P Linear Discovery (LD) Message

The New BN Declaration (NBD) Algorithm - Initialization Two types of nodes : l Backbone Nodes ( BNs ): part of the l backbone. Can be used for routing , and other functions (data compression, etc.) Non-Backbone Nodes ( NBs ): not part of 2 l F 3 3 2 2 G V W 2 Z backbone. Can perform basic sensing 2 X NBD(C, CEG, 3) NBD(C, CE, 2) NBD(K, KJS, 3) NBD(C, CE, 2) S operation. NBD(M, MU, 2) E 1 NBD(M, MU, 2) 1 1 U NBD(K, KJ, 2) H 1 T J 1 NBD(B, B, 1) NB nodes need to find paths to nearest BN nodes NBD(L, L, 1) NBD(C, C, 1) NBD(C, CH, 2) l Sink NBD(C, C, 1) NBD(M, M, 1) Discovery Initiator Node NBD(K, K, 1) to use them for routing. C A B L K I M The newly discovered BN nodes will broadcast NBD(B, B, 1) NBD(K, K, 1) l 1 NBD(I, I, 1) Q 1 Y a New BN Declaration ( NBD ) message to NBD(K, KY, 2) N NBD(B, BQ, 2) NBD(K, KY, 2) accomplish this task. 1 2 R NBD(I, IN, 2) O 2 NBD(K, KYP, 3) 2 NBD message has the following fields: P l Link in Discovered Backbone Backbone Node (BN) messageID : to prevent looping l Link Outside Backbone Non-Backbone Node (NB) NB Discovery (NBD) Message Link Outside Backbone sourceBNID : ID of BN node l myID : ID of forwarding node l BNDRingSize : size of broadcast ring ρ l numOfHops : traversed number of hops from BN l node PATH_to_BN : accumulated path to BN node l

NBD Message Propagation in New BN Node Discovery Algorithm 2 F 3 3 2 2 G V W Z 2 X 2 NBD(C, CEG, 3) NBD(C, CE, 2) NBD(K, KJS, 3) S NBD(C, CE, 2) NBD(M, MU, 2) E 1 NBD(M, MU, 2) 1 1 U NBD(K, KJ, 2) H 1 J 1 T NBD(B, B, 1) NBD(L, L, 1) NBD(C, C, 1) NBD(C, CH, 2) Sink Discovery Initiator Node NBD(C, C, 1) NBD(M, M, 1) NBD(K, K, 1) C A B L K I M NBD(K, K, 1) NBD(B, B, 1) 1 NBD(I, I, 1) Q 1 Y NBD(K, KY, 2) N NBD(B, BQ, 2) NBD(K, KY, 2) 1 2 R NBD(I, IN, 2) 2 O NBD(K, KYP, 3) 2 P Link in Discovered Backbone Backbone Node (BN) Link Outside Backbone Non-Backbone Node (NB) NB Discovery (NBD) Message Link Outside Backbone

The LNBN and L2BN Algorithms l Two metrics l Consider thick LSN with: Number of generated messages L: length l l for discovery T: thickness l Average number of hops for each l l Requires four anchor nodes SN to send messages to the sink I: the discovery initiator l l LNBN : does not explicitly minimize S: the sink l the number of hops to the sink Two other anchor nodes: l l Flooding can be used to minimize U(L/2, T/4) l the number of hops . Each SN send V(L/2, 3T/4) l LD LD message to sink . Extreme l With upper and lower case paths SNs in the upper l L2BN balances the two strategies. and lower regions have shorter path to sink l Discover backbone with two paths using anchor nodes in the middle .

Note these two paths are not necessarily node disjoint

Simulation Simulation to validate and evaluate the l algorithms. Thick LSN generated according to model . l Modeled as rectangle in our simulation l Key parameters : l Thickness of LSN (i.e. width): W – Set to l 500 m. Length of LSN: L – set to 10000 m. l Number of sensor nodes: N – Set to 1000 l Node communication range : Range – Set to l 100 m. Position of each sensor node uniformly l generated within 2-dimensional rectangle Initiator node is leftmost node in 2-D l rectangle Sink is rightmost node l Illustration of the backbone path where W = 500, L = 2500, N = 300, and Range = 200. Performance metrics : l Time for backbone discovery l Number of LD and SF messages used in l discovery Number of new backbone declaration ( NBD ) l messages

LNBN on large instances When number of SNs increases , backbone discovery time increases. l Number of LD+SF messages increases as number of SNs increases l increasingly, the increasing speed also increases: messages are spread in a broadcast nature, so messages increase proportionally to the square of the number of SNs.

Comparison results of LNBN and L2BN Comparison of LNBN and L2BN under varying number of nodes while fixing the range at 100 Comparison of LNBN and L2BN under varying range while fixing the number of nodes at 1000

Average No. of Hops and Total No. of Message Forwardings When number of normal data messages exceeds 2,000 , total number of l message forwardings in L2BN becomes less than that of LNBN. Since number of SNs in the WSN of interest typical exceeds 1,000 , the l number of normal data messages can easily exceed 2,000 .