1 B-MAC Implementation B-MAC Implementation Low Power Listening - - PDF document

1

Versatile Low Power Media Access for Wireless Sensor Networks

Presented By: Eitan Marder-Eppstein

Presentation Outline

Introduction B-MAC Implementation B-MAC Results Critique Comparison

Why is B-MAC Needed?

S-MAC and T-MAC:

No flexibility – What if network conditions change? Too much code – Limited memory on motes Doesn't scale well – As network size increases, nodes must maintain more schedules

Well... there's also WiseMAC:

Has no mechanism to reconfigure based on services using the protocol

Motivated by needs of monitoring applications

B-MAC's Idea of MAC for WSN

B-MAC: A carrier sense media access protocol for wireless sensor networks B-MAC Goals:

Low Power Operation Effective Collision Avoidance Simple Implementation, Small Code and RAM Size Efficient Channel Utilization Regardless of Data Rate Reconfigurable by Network Protocols Tolerant to Changing Network Conditions Highly Scalable

B-MAC Implementation

Carrier Sense Multiple Access with Collision Avoidance

Listen for a pre-determined amount of time for network activity If the network is idle then send a packet If activity is detected then wait for a random amount of time (called backoff factor) before re-attempting transmission Backoff counter only decremented when channel is clear

A B C D E Sender Receiver Backoff Set Backoff Set 5 F Hidden Terminal Could Exist ,4,3 2,1,0

B-MAC Implementation

Clear Channel Assessment (CCA)

Ambient noise changes depending on environment B-MAC employs software to estimate the noise floor Search for outliers significantly below the noise floor since a valid packet could never have one If clear then transmit If busy then backoff

2

Low Power Listening (LPL)

When a node awakes it checks for radio activity using CCA If activity is detected, stay awake to receive the packet and then sleep If no packet (false positive) timeout forces node to sleep

Preamble Length

Preamble must be at least as long as duty cycle for reliable data reception

B-MAC Implementation

B A Sender Receiver

100ms Preamble 100ms Duty Cycle

B-MAC Implementation

B-MAC = Link Protocol

Only a small core of media access functionality included with B-MAC Network services (organization, synchronization, routing) built above B-MAC

What about Hidden Terminal, Fragmentation, etc?

No “built in” mechanisms for handling these Instead, B-MAC exposes a set of interfaces to allow services to tune B-MAC Protocols built on B-MAC can optimize performance in their environment

Interface MacControl{ EnableCCA();/DisableCCA(); EnableAck();/DisableAck(); HaltTx(); } Interface MacBackoff{ initial/congestionBackoff(); } Interface LowPowerListening{ Set/GetListeningMode(); Set/GetTransmitMode(); Set/GetPreambleLength(); Set/GetCheckInterval(); }

B-MAC Implementation

Link Layer Acknowledgment

If enabled, B-MAC transfers an ack code after receiving a unicast packet

Small Code Size

ROM and RAM limited Small code size important

B-MAC – Adjusting Parameters

Calculate Optimal Parameters

Specifically, minimize a node's energy consumption to maximize lifetime Solve system of 6 equations to find minimum energy consumption for a given network configuration:

B-MAC – Results

Node lifetime in years based on LPL check time and network density If both neighbourhood size and check interval are known, their intersection gives expected lifetime using optimal parameters How duty cycle is affected by network density and LPL interval Best check interval is lowest line at a given network density

B-MAC – Results

Shows trade off of more frequently checking the radio to shorten packet transmission time Penalty for idle listening is much more severe than sending packets that are longer than necessary

3

B-MAC – Protocol Comparison Results - Latency/Energy

Critique

B-MAC to WiseMAC Comparison

After talk of WiseMAC as the only serious competitor to B-MAC at the release date of this paper, the authors neglect to test against it.

Is scheduling always a bad thing

B-MAC does away with synchronization between nodes, but does this always improve performance. Is there a better way to schedule.

Energy Floor for LPL

The authors of the paper mention that there is an energy floor for LPL. Doesn’t this mean that for a network with high latency constraints B-MAC is not the most energy efficient protocol. In fact, results from the paper confirm that S-MAC uses less energy than B-MAC at high latency.

Optimizing for Average Conditions: What if they often change?

Comparison and Questions

SCP

Synchronize the entire network in an effective and efficient manner Eliminate the long preamble associated with LPL Adapts well to variable traffic loads Reduce duty cycle from 1-2% for current MAC protocols to 0.1%

?

B-MAC – Protocol Comparison Results - Throughput

B-MAC – Protocol Comparison Results - Fragmentation