Cross-Layer Application-Specific WSN Design over SS-Trees

-Prepared by Amy

Outline

• Background Introduction

• Sleep Scheduling Issues & the SS-Tree Concept

• SS-Tree Operational Stages

• SS-Tree Computation

• SS-Tree Operational Specifics & Sleep Scheduling

• Conclusions and Future Work

Background Introduction

• Wide-area surveillance WSN

applications−expected lifetime −limited battery supply

• Energy Efficiency is paramount

• Adaptive sleep schedules to

minimize energy lost

Background Introduction

• Sleep scheduling: −shorten the time radio transceiver engaged in

idle listening

• Good impact:−reduced overhearing

• Ensuing problem: − link table entries expire prematurely−control and data packet compete for resources−real-time data reporting function reduced

Background Introduction

• Ultimate Design Goal:−Balance:

• sensing requirements• end-to-end data communication overhead• network control effectiveness

−With energy efficiency−Through a cross-layer sleep scheduling

scheme

Sleep Scheduling Issues

• Not recommended:

• Random sleep scheduling−detrimental effect on network connectivity

and topology control efficiency

• Global sleep scheduling−network-wide communication blackout

• Groups of leaf nodes sleep scheduling−non-leaf nodes depleting battery reserves

sooner

Sleep Scheduling Issues

• Using coordinated sleep scheduling−Realize the benefits:

• reduced overhearing • reduced packet collision• simplified topology

−Without sacrifice:• network connectivity • sensing capabilities

SS-Tree Concept

SS-Tree Concept

• Advantages:−Avoid overburdening any set of nodes

from being the sole virtual backbone−Increase monitoring sensitivity (greater

event reporting windows) without altering communication duty cycle(reporting frequencies)

SS-Tree Concept--issues to be considered Gaps appearing in

between the active period of adjacent SS-Tree

SS-Tree Concept--issues to be considered -- Blackout duration -- Sleep period -- number of mutually adjacent SS-Trees -- Active period

Number of distinct live pathTo guarantee 100% real-time event reporting capability

Not feasible due to limited nodal densityAnd high SS-Tree computation complexityNot necessary to approach real-timeIntuition suggests the number of SS-TreeShould less than the average nodal degree

SS-Tree Concept--issues to be considered

Drawback: timer-drivenData cannot be simultaneouslyGathered from all SS-Trees

SS-Tree Operational Stages

SS-Tree Operational Stages

• Network Initialization: − gather network connectivity information, − compute the SS-Trees− disseminate the sleep schedules

• Sleep: − shut down the radio transceiver− processor and sensing unit remain active

• Hibernation: − Shutting down all hardware components − except for a tiny low-power wakeup timer

SS-Tree Operational Stages

• Active: −all data reporting −network maintenance tasks are performed

• Failure Recovery: −data sink repair or reconstruct SS-Trees

• Neighborhood Update: −neighboring nodes exchange local information −for each other’s sleep schedule

SS-Tree Computation

SS-Tree Computation

• A greedy depth-first approach

• From the bottom-up on a branch-by-

branch basis

• Proceeds in a number of iterations

• In each iteration an end-to-end minimum

cost path is appended to one of the SS-

Trees.

SS-Tree Computation

SS-Tree Computation

SS-Tree Computation

SS-Tree Computation

SS-Tree Operational Specifics & Sleep Scheduling

• Major task – determine an optimal sleep

schedule that maximizes energy efficiency

• Short active period -> high transmission latency

• Longer active period -> increase sleep time

between two consecutive active periods

• Determine an upper bound of active period− balance low communication duty cycle− monitoring sensitivity− end-to-end packet transmissions

SS-Tree Operational Specifics & Sleep Scheduling

Network Layer Routing

SS-Tree Operational Specifics & Sleep Scheduling

• Some flexible strategies in manipulating application requirements:−Compact query formats

• shrink packet size by formatting data types • reduce hop-by-hop transmission time

−Aggressive data aggregation • duplicate suppression • reduce unnecessary packet exchange

−Hop-by-hop ACK in MAC layer • instead of end-to end ACK in transport layer • reduce energy expenditure

SS-Tree Operational Specifics & Sleep Scheduling

SS-Tree Operational Specifics & Sleep Scheduling

• Medium Access Control−Prefer single-channel unslotted CSMA

• simplicity• greater scalability• looser time synchronization requirements

−Bypass the RTS/CTS handshake • long end-to-end propagation delay

SS-Tree Operational Specifics & Sleep Scheduling

Timing components constituting a single active period

Round-trip time recorded for node I on its respective SS-Tree

SS-Tree Operational Specifics & Sleep Scheduling

SS-Tree Operational Specifics & Sleep Scheduling

SS-Tree Operational Specifics & Sleep Scheduling

SS-Tree Operational Specifics & Sleep Scheduling

IACK works better in reducing the time when the size of C/D packet is comparable to that of EACK

Conclusion and Future Work• Following issues will be explored:

1. For a given random topology, what is the maximum number of SS-Trees that can be constructed to minimize the number of shared nodes?

2. For a given number of nodes, what is the optimal method of deployment that ensures 100% coverage of the subject area while maximizing the number of available SS-Trees with minimum shared nodes?

3. What are the suitable neighborhood discovery and failure recovery strategies for the SS-Tree design?

The End