ST-MAC: Spatial-Temporal MAC Scheduling for Underwater Sensor Networks

Chih-Cheng Hsu, Kuang-Fu Lai, Cheng-Fu Chou, Kate Ching-Ju Lin

IEEE INFOCOM 2009

Outline Introduction Related Work ST-MAC Framework Performance Evaluation Conclusion

Introduction Similar to terrestrial sensors, energy efficienc

y is critical considerations in UWSNs unlike terrestrial sensor utilize acoustic waves propagation is slower than RF

In UWSNs, must consider the locations of the receiver and potential interferers “Spatial-Temporal Uncertainty”

Spatial-Temporal Uncertainty

TDMA-based MAC protocols To utilize time slots efficiently, the vertex

coloring scheme is used for scheduling Propose a novel heuristic algorithm, called

Traffic-based + One-Step Trial Approach (TOTA)

Model the scheduling problem as a Mixed Integer Linear Programming (MILP) model

ST-MAC Framework Most of underwater sensors are deployed to

get data of interest periodically Each node can estimate signal-to-noise-ratio

determining interference relationships measuring the propagation delay

ST-MAC is to compute the schedule each sensor nodes knows when to switch to

sleeping mode

ST-GG Construction Base station can acquire the routing topology

G(V,E), V is a set of sensors E denotes a set of transmission links

Define PD(vi, vj) as the propagation delay between node vi and vj

Spatial-Temporal Conflict Graph Spatial-Temporal Conflict Graph (ST-CG), a

directed graph G(V ,E) V = E and E is the set of conflict relationships bet

ween any two transmissions Conflict relation Conflict(u → v),

exists if transmission of link u affects reception of link v

Example of Conflict

Two Links With Common Node Case 1.1: u.dst = v.dst

Case 1.2: u.src = v.src

Case 1.3: u.src = v.dst

Two Links Without Common Nodes

Two Links Without Common Nodes Case 2.1: ONLY one of INTER(u, v) and INT

ER(v, u) is TRUE cc,d = −3

Case 2.2: both INTER(u, v) and INTER(v, u) are TRUE conflict delays cb,c = −4 and cc,b = −2

Traffic-based One-step Trial ApproachS

MReal

MTest

Traffic-based One-step Trial Approach

MReal

MTest

S

Theoretical Analysis Propose mixed Integer Linear Programming

model solve the new type of the vertex-coloring problem

in ST-CG optimally as a benchmark to quantitatively evaluate the

performance of existing heuristic methods

Propagation Delay Constraint

Modified equations by using the Big-M method

binary variable used to transform disjunction

Inter-frame Constraint Transmission of link j in next frame must not

conflict with the reception of link I

Transmission of link i in the next frame must not conflict with the reception of link j

Minimize Problem

Performance Evaluation All simulations are implemented in NS2 Two different scales

the small topology case: 6 - 12 nodes the large-topology case: 81 - 144 nodes

Small Central-Sink Topology

Large Central-Sink Topology

Large Cluster-Sink Topology

Energy Cost

Unknown Traffic Scenarios

Conclusion Proposes a TDMA-based scheduling to solve

Spatial-Temporal Uncertainty in UWSNs Construct ST-CG that includes the propagatio

n delay information present TOTA, to solve more effectively

Derive a MILP formulation solving the optimal solution of the vertex-colorin

g problem in ST-CG graph