Topic/project/effort description

1
Topic/project/effort description
First item planned. Add more text as
necessary. Second item planned. Add more text
as necessary. Add other points as necessary

• MAIN ACHIEVEMENT
• Placeholder explanatory text. Replace with text
  and diagrams as necessary.
• HOW IT WORKS
• Placeholder explanatory text paragraph. Replace
  with text and diagrams as necessary.
• ASSUMPTIONS AND LIMITATIONS
• Limitation or assumption
• Another limitation or assumption

Primary answer here. Add more text as
necessary. First bullet point Additional as
necessary
First key insight. Add more text as
necessary. Second key insight. Add more text as
necessary. Add other points as necessary
Primary answer here. Add more text as
necessary. First key point Additional as
necessary
A sentence why it is important/useful
2
Network coding (NC) for efficiency robustness
Analytical model of NC static multicast scenario
shows superior goodput and graceful degradation
with packet loss

• MAIN RESULT
• Analyzed and implemented network coding
  algorithms for dynamic wireless networks.
• HOW IT WORKS
• Topology information is collected to compute
  subgraphs. Source nodes mix packets which
  forwarded by subgraph nodes to unicast or
  multicast destinations.
• ASSUMPTIONS AND LIMITATIONS
• Needs further integration with reliable hyperlink
  protocol.
• Needs further integration with channel access
  protocol.
• Control overhead for baseline and CONCERTO
  protocols not included in analysis.
• Potential additional gains from inter-session
  coding not included in analysis.

20
Bandwidth savings ratio (BSR)
Bandwidth
Savings
Ratio
15
Phase 2 BSR Target
Goodput
Network Coding
10
NORM
Multicast ARQ
Unicast ARQ

Network Coding
NORM
5
5
Phase 1 BSR Target
Multicast ARQ
Traditional packet copying (C) and forwarding (F)
is inefficient and fails to exploit the
availability of inexpensive memory and CPU
resources.
Unicast ARQ

0 20 40
60 80 100
Probability of Loss
Network coding moves information rather than
packets. It exploits computing (?) and storage (
) to provide robust performance in degraded and
congested settings.
Demonstrate 10x bandwidth reduction compared to
baseline MANET implementation using realistic
scenario and traffic load
Analysis indicates potential to meet Phase 1
metrics. Partial network stack demonstrated.
3
Network coding as a unifying architectural
principle
Philosophy of network coding as infrastructure
reduces number of protocols dramatically,
simplifying configuration and algorithm
development.

• MAIN RESULT
• Simplified network stack architecture based on
  coding
• HOW IT WORKS
• Unicast, broadcast and multiple-path routing
  are special cases of multicast subgraphs.
  Rateless coding integrates packet level FEC and
  ARQ.
• ASSUMPTIONS AND LIMITATIONS
• Analyzed, but have not implemented,
  network-coding compatible backpressure,
  admissions control and rate control algorithms.

Existing protocols were developed to solve
specific problems (unicast, multicast, link level
reliability, end-to-end reliability) and do not
form a cohesive whole.
Network coding subsumes unicast, multicast,
multiple path routing, opportunistic routing,
packet level FEC, ARQ and rateless coding.
Incorporate intra-session coding. Demonstrate
that multiple protocols can be replaced with
network coding.
CONCERTOs network coding approach simplifies
MANET architecture
4
Progress on a backpressure-informed media access
control
TDMA-based protocols require close coordination
and tight time sync to achieve optimal channel
utilization Random access approaches are
simple, but have poor utilization
RC-MAC in MARCONI achieves near-optimal channel
access without TDMA overhead.

• MAIN RESULT
• Implemented and demonstrated differentiated
  random access with backpressure signaling
• HOW IT WORKS
• Basic RC-MAC Nodes with highest urgency have
  highest channel access probability
• MARCONI RC-MAC Normalized backpressure signals
  specify max urgency wi of each node
• P(access) ? 1 for most urgent nodesP(access) ? 0
  for least urgent nodes
• ASSUMPTIONS AND LIMITATIONS
• Assumes queue length metric includes all criteria
  that determine message urgency

2-user shared medium
Ideal TDMA
user 2 access time
Current RC-MAC
user 1 accesstime
State of the art (802.11) on small packets (e.g.
VoIP
Backpressure congestion signal specifies urgency
of channel access across nodes, not just within a
node

• Refine urgency weighting function for
  delay-sensitive traffic
• Refine urgency vs. fairness tradeoff

Our regulated contention MAC approaches optimal
channel utilization without the overhead of TDMA
5
Progress on Joint Routing and Admission Control
In severely challenged networks, admission
control rejects some flows to guarantee QoS of
others, improving overall delivery of bulk files
(green vs. yellow) and streaming video (blue vs.
red)

• MAIN RESULT
• Implemented route discovery and admission
  control protocol that tests for flow feasibility
  and decides feasibility using backpressure signal
  
• HOW IT WORKS
• Forward sweep (join query) identifies possible
  paths to destination
• Return sweep (join reply) rejects infeasible
  paths, choosing one with greatest surplus
  capacity
• Flows admitted only after route discovery
  identifies a path with sufficient resources
• ASSUMPTIONS AND LIMITATIONS
• Problem formulation collapses all capacity and
  QoS into a scalar routing metric
• Current design implementation unicast only

• Ad hoc routing metrics may not match network
  goals
• Resource reservation infeasible for MANETs
• Route discovery does not check that network can
  support new traffic

Backpressure complements channel utilization and
link capacity in determining the feasibility and
admissibility of a new route
Flow rejected unless capacity exists and
congestion is feasible

• Implement multicast routing
• Improve route adaptation to manage changes in
  MANET dynamics and account for network-wide
  impact of flows

Our JRAP protocol aligns routing and admission
goals with optimal control objectives