Cross-Layer Design:
A Case for Standardization
Presented by
Shreya Sarker
st123286
Presentation Outline
INTRODUCTION
BACKGROUND
CROSS LAYER FRAMEWORKS (CLF)
REVIEW OF CLF PROPOSALS
METHODOLOGY (CROSS-LAYER DESIGN BY
EXAMPLE)
• REVIEW
• CONCLUSIONS
•
•
•
•
•
INTRODUCTION
• Wireless Sensor Networks (WSNs)
• Open System Interconnection (OSI) reference
model
• Cross-layer design (CLD)
• Cross-Layer Frameworks
• Standardized?
• Summary/Objective
BACKGROUND
• The OSI Model
BACKGROUND
• Cross-Layer Design
 Classes
BACKGROUND
• Potential issues with CLD
I. Unintended Cross-Layer Interaction: lead to
unforeseen dependencies which are not predicted
by simulation.
II. Stability: a given CLD may make changes at one
layer based upon feedback from another layer.
III. Long-term Sustainability: a change made at any
given layer could affect the functionality of any
other layer.
CROSS LAYER FRAMEWORKS (CLF)
1. Direct Inter-layer Communication
2. Shared Super Layer
3. New Abstractions
CROSS LAYER FRAMEWORKS (CLF)
1. Direct Inter-layer Communication: The DynamicMulti-Attribute Cross-Layer Design (DMA-CLD)
framework was proposed.
But, direct inter-layer communication is not a good
approach for CLD.
CROSS LAYER FRAMEWORKS (CLF)
2. Shared Super Layer: The Optimization Agent
(OA) Framework is proposed.
But, giving the OA unfettered access to every layer is
dangerous.
CROSS LAYER FRAMEWORKS (CLF)
3. New Abstractions: The TinyCubus framework is
described.
CROSS LAYER FRAMEWORKS (CLF)
3. New Abstractions:
Divides WSN into three interdependent frameworks:
I.
Data Management Framework: selects
appropriate operating modes for each sensor
node.
II. Cross-Layer Framework: incorporates sets of
communication layers in order to create a
functional communication system.
III. Configuration Engine: responsible for network
configuration and node updating.
CROSS LAYER FRAMEWORKS (CLF)
3. New Abstractions
But, It is not clear how a variety of CLD proposals
would be cleanly incorporated into this framework,
nor exactly how the application requirements and
optimization goals should be chosen.
REVIEW OF CLF PROPOSALS
• STANDARD???
• For the creation of a universal CLF standard, more
work and collaboration in this area is required.
Additionally, a good CLF will define more than just
communication blocks and allowable interactions.
• Joint effort will be required in order to produce a
capable CLF.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
Evaluation Criterion (Forms of application
Requirement)
a. Optimization objectives: set of network lifetime
b. System constraints
 Constructive constraints or
 Destructive constraints
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
Evaluation Criterion
a.
b.
c.
d.
Define the involved layers
Check system model and assumptions
State the Optimization Objectives
State the system constraints, constructive and
destructive
e. Explain the nature of the optimization
f. Define new requirements for each involved layer
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
a. Involved Layers: Physical, MAC, and Network
b. Model & Assumptions:
 Assume a time-invariant network topology
described by a directed graph G = (V, L)
•
•
V is the set of nodes
L is the set of links
 Assume Single commodity flow
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
c.



Optimization Objectives
Total Power Consumption: Minimize
Minimum Node Lifetime: Maximize
Concave Functions of Node Lifetimes: as long as
one node in that block can sense and report
data
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
d.



System Constraints
Data Rate: Destructive
Energy Consumption: Destructive
Flow Conservation: Destructive
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
e. Optimization Methodology
 First, consider the TDMA case for lifetime
maximization, subject to flow conservation, energy
conservation, and other constraints.
 Second, the total network power minimization
scheme is considered.
 Third, case for a slotted network with interfering
links is studied.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
1. Proposal: Optimization of Transmission
Schemes
f. Cross-Layer Method & Interfaces
 There is Dynamic vertical calibration between the
Physical, MAC, and Network layers, where the
central controller is responsible for updating the
calibration of every node.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
a. Involved Layers: MAC, and Network
b. Model & Assumptions:
 Assume sensor network model is defined as an
undirected graph G = (V, E)
• V is the set of sensor nodes
• E is the set of undirected wireless links between
nodes.
 To pass information between cluster-heads, they
identify and use neighboring members from
different clusters to relay information.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
c. Optimization Objectives
 Energy efficiency such that minimum node
lifetime is maximized.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
d. System Constraints (For Network and MAC Layer)
 End-to-end delay (Destructive Constraint)
Where,
Delay constraint τ
Drep is defined as the total data reporting time
Dprop is the one-hop propagation delay.
Dproc is the amount of processing time for
data fusion per cluster head.
Hint is the number of intermediary clusterheads.
Finally, k is the number of nodes within the
data source cluster.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
d. System Constraints
 Spatial Correlation (Constructive Constraint): The
vector of measurements from any two nodes can
be used to calculate a spatial correlation
coefficient, ρij.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
2. Proposal: Adaptive Quality of Service
f. Cross-Layer Method & Interfaces
 For the routing scheme, new interfaces into the
Network layer are required, such that an optimal
route to the data sink can be chosen.
 For the MAC scheme, the application layer is
designed to specifically interact with a TDMA
MAC, such that slots for a single node-per-block
can be chosen.
METHODOLOGY (CROSS-LAYER
DESIGN BY EXAMPLE)
Review
No common methodology
observed in developing or
evaluating these ideas.
CONCLUSIONS
• OSI approach to communication systems
design is not optimal.
• Two proposed approaches for a CLF are
viable but lack enough specification and
definition.
• STANDARDIZED FRAMEWORK FOR CLD
NEEDED