Adaptation of TDMA Parameters Based on Network Conditions Bora Karaoglu Tolga Numanoglu Wendi Heinzelman Department of Electrical and Computer Engineering.

Adaptation of TDMA Parameters Adaptation of TDMA Parameters Based onBased on

Network ConditionsNetwork Conditions

Adaptation of TDMA Parameters Adaptation of TDMA Parameters Based onBased on

Network ConditionsNetwork Conditions

Bora Karaoglu Tolga Numanoglu Wendi Heinzelman

Department of Electrical and Computer Engineering

University of Rochester, NY, USA

Bora Karaoglu Tolga Numanoglu Wendi Heinzelman

Department of Electrical and Computer Engineering

University of Rochester, NY, USA

MotivationMotivation Capacity Each tx occupies some

part of the capacity

Capacity Each tx occupies some

part of the capacity

MotivationMotivation Clustering approach:

Divide into a number of chunks

CHs use chunks

Question? How many chunks?

Work summarized in: Analytical model Optimization

Clustering approach: Divide into a number of

chunks CHs use chunks

Question? How many chunks?

Work summarized in: Analytical model Optimization

AgendaAgenda

Protocol Overview: MH-TRACE Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Protocol Overview: MH-TRACE Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Multi-Hop Time Reservation Using Adaptive

Control for Energy Efficiency Multi-Hop Time Reservation Using Adaptive

Control for Energy Efficiency

TDMA Soft clustering

CHs responsible for channel access only

Inter cluster communication is allowed

TDMA Soft clustering

CHs responsible for channel access only

Inter cluster communication is allowed

Nf = 6

Protocol Overview: MH-TRACEProtocol Overview: MH-TRACE

Factors limiting performance: Dropped Packets

Real-time communication Limited Local Capacity

Clustering Uneven distribution of Load

Node Distributions Mobility

Collisions Spatial Reuse

Limited capacity Divisions

Factors limiting performance: Dropped Packets

Real-time communication Limited Local Capacity

Clustering Uneven distribution of Load

Node Distributions Mobility

Collisions Spatial Reuse

Limited capacity Divisions

AgendaAgenda

Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Analytical ModelAnalytical Model

Shortcomings of Simulations Substantial Processing Power and Time

Repetitions for statistical accuracy Valid only for the parameters set used

Scalability of Simulation Area Edge Effects

Shortcomings of Simulations Substantial Processing Power and Time

Repetitions for statistical accuracy Valid only for the parameters set used

Scalability of Simulation Area Edge Effects

Dropped PacketsDropped Packets

Probability of Dropping a Packet

Capacity per Cluster: Number of Data Slot per Frame

Nonlinear relation between Load and Pdp

Detailed probability distribution of Load is needed

Probability of Dropping a Packet

Capacity per Cluster: Number of Data Slot per Frame

Nonlinear relation between Load and Pdp

Detailed probability distribution of Load is needed

Dropped PacketsDropped Packets

Ps: Ratio of number of nodes in spurt to all nodes Voice Activity Detector

NCH: Number of CHs each node can receive access from NCM: Number of nodes in the Cluster

Ps: Ratio of number of nodes in spurt to all nodes Voice Activity Detector

NCH: Number of CHs each node can receive access from NCM: Number of nodes in the Cluster

Dropped PacketsDropped Packets

Effect of Dropped Packets on Throughput Considering Rx Throughput

Each node all one hop neighbors

Effect of Dropped Packets on Throughput Considering Rx Throughput

Each node all one hop neighbors

CollisionsCollisions

Number of frames (Nf) vs. co-frame CH separation

(dch) Labeling structure used in cellular systems

Co-frame CH separation (dch) vs. number of collisions (fcoll)

Correlation between Number of Nodes that can cause collisions Number of Collisions

Number of frames (Nf) vs. co-frame CH separation

(dch) Labeling structure used in cellular systems

Co-frame CH separation (dch) vs. number of collisions (fcoll)

Correlation between Number of Nodes that can cause collisions Number of Collisions

AgendaAgenda

Soft Clustering Approaches Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Soft Clustering Approaches Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Proof of ConceptProof of Concept

Number of Packets Lost per Superframe (Nf = 6) Number of Packets Lost per Superframe (Nf = 6)

Proof of ConceptProof of Concept

Number of Packets Lost per Superframe (Nf = 8) Number of Packets Lost per Superframe (Nf = 8)

AgendaAgenda

Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Protocol Overview Analytical Model

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Optimization of TDMA parametersOptimization of TDMA parameters optimization with corresponding throughput

figures with respect to the maximum realizable throughput

optimization with corresponding throughput figures with respect to the maximum realizable throughput

Theoretically Optimized Nf

Conclusions and Future WorkConclusions and Future Work The model

Accurate Can be used in optimization of parameters Instantaneous results for changing

Transmission Power Propagation Model PHY Specs: Freq, Threshold values …

Asymptotic behavior Energy consumption

Average node sleep/awake durations Average energy consumption per node Node and CH comparison wrt energy consumption Optimization of Nf wrt energy consumption

We are going to add effects of upper layers into the model

The model Accurate Can be used in optimization of parameters Instantaneous results for changing

Transmission Power Propagation Model PHY Specs: Freq, Threshold values …

Asymptotic behavior Energy consumption

Average node sleep/awake durations Average energy consumption per node Node and CH comparison wrt energy consumption Optimization of Nf wrt energy consumption

We are going to add effects of upper layers into the model

Thanks!Questions&Comments?

Contact Info:

Web : www.ece.rochester.edu/~karaoglu/

E-mail : [email protected]

Thanks!Questions&Comments?

Contact Info:

Web : www.ece.rochester.edu/~karaoglu/

E-mail : [email protected]

PHY Layer AbstractionPHY Layer Abstraction

BW Each tx occupies some

part of the BW Transmissions should

overcome any noise present in the space of the BW

Divide Spatial reuse

BW Each tx occupies some

part of the BW Transmissions should

overcome any noise present in the space of the BW

Divide Spatial reuse

PHY Layer AbstractionPHY Layer Abstraction

TDMA: Divide BW along time

axis

Clustering: Distribute parts of BW

spatially among clusters

TDMA: Divide BW along time

axis

Clustering: Distribute parts of BW

spatially among clusters

Protocol OverviewProtocol Overview

TDMA Soft Clustering

CHs responsible for channel access only

Inter cluster communication is allowed

TDMA Soft Clustering

CHs responsible for channel access only

Inter cluster communication is allowed

Analytical AnalysisAnalytical Analysis

Shortcomings of Simulations Substantial Processing Power and Time

Repetitions for statistical accuracy Valid only for the parameters set used

Scalability of Simulation Area Edge Effects

Shortcomings of Simulations Substantial Processing Power and Time

Repetitions for statistical accuracy Valid only for the parameters set used

Scalability of Simulation Area Edge Effects

Analytical AnalysisAnalytical Analysis

Factors limiting performance: Dropped Packets

Real-time communication Limited Local Capacity

Clustering Uneven distribution of Load

Node Distributions Mobility

Collisions Spatial Reuse

Limited BW Divisions

Factors limiting performance: Dropped Packets

Real-time communication Limited Local Capacity

Clustering Uneven distribution of Load

Node Distributions Mobility

Collisions Spatial Reuse

Limited BW Divisions

Dropped PacketsDropped Packets

Probability of Dropping a Packet

Capacity per Cluster: Number of Data Slot per Frame

Nonlinear relation between Load and Pdp

Detailed probability distribution of Load is needed

Probability of Dropping a Packet

Capacity per Cluster: Number of Data Slot per Frame

Nonlinear relation between Load and Pdp

Detailed probability distribution of Load is needed

Dropped PacketsDropped Packets

ps: Probability of a node to be in spurt duration

pA: Probability of a node to be in the communication range of a CH

pd: Probability of a node that is in the communication range of a CH to choose that CH as its channel access provider Independent of Node Density assumed constant

ps: Probability of a node to be in spurt duration

pA: Probability of a node to be in the communication range of a CH

pd: Probability of a node that is in the communication range of a CH to choose that CH as its channel access provider Independent of Node Density assumed constant

Dropped PacketsDropped Packets

ps: Probability of a node to be in spurt duration

pA: Probability of a node to be in the communication range of a CH

pd: Probability of a node that is in the communication range of a CH

to choose that CH as its channel access provider

pdn = ps pA pd

ps: Probability of a node to be in spurt duration

pA: Probability of a node to be in the communication range of a CH

pd: Probability of a node that is in the communication range of a CH

to choose that CH as its channel access provider

pdn = ps pA pd

CollisionsCollisions

Number of frames (Nf) vs. co-frame CH

separation(dch) Labeling structure used in cellular systems

Number of frames (Nf) vs. co-frame CH

separation(dch) Labeling structure used in cellular systems

r

r

r

CollisionsCollisions

co-frame CH separation (dch) vs. number of collisions

(fcoll)

co-frame CH separation (dch) vs. number of collisions

(fcoll)

CollisionsCollisions

co-frame CH separation (dch) vs. number of collisions (fcoll)

co-frame CH separation (dch) vs. number of collisions (fcoll)

NnCH : Expected number of nodes in the cluster

Nn : Total number of nodes

NC : Number of cluster in 2*rcomm range

V : Region bounded by the circle with radius 2*rcomm around origin

fcoll : number of packets lost per SF due to collision

AgendaAgenda

Soft Clustering Approaches Protocol Overview Analytical Analysis

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Soft Clustering Approaches Protocol Overview Analytical Analysis

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Proof of ConceptProof of Concept

Total Number of Packets Lost per Superframe Total Number of Packets Lost per Superframe

Proof of ConceptProof of Concept

RX Throughput per Superframe RX Throughput per Superframe

AgendaAgenda

Soft Clustering Approaches Protocol Overview Analytical Analysis

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Soft Clustering Approaches Protocol Overview Analytical Analysis

Dropped Packets Collisions

Proof of Concept Optimization of TDMA parameters

Optimization of TDMA parametersOptimization of TDMA parameters

Other Uses of the ModelOther Uses of the Model

Instantaneous Analysis Results for changing Transmission Power Propagation Model PHY Specs: Freq, Threshold values …

Asymptotic Behavior Energy Consumption

Average node sleep/awake durations Average energy consumption per node Node and CH comparison wrt energy consumption Optimization of Nf wrt energy consumption

Instantaneous Analysis Results for changing Transmission Power Propagation Model PHY Specs: Freq, Threshold values …

Asymptotic Behavior Energy Consumption

Average node sleep/awake durations Average energy consumption per node Node and CH comparison wrt energy consumption Optimization of Nf wrt energy consumption

Throughput Per NodeThroughput Per Node

Energy Consumption per NodeEnergy Consumption per Node

Thanks!Questions&Comments?

Thanks!Questions&Comments?