Title: Cross Layer Architectures for Tactical Ad Hoc Networks equipped with Space-Time Code Processing Capabilities
1Cross Layer Architectures for Tactical Ad Hoc
Networks equipped with Space-Time Code Processing
Capabilities
Presented by Prof. J.J. Garcia-Luna-Aceves UC
Santa Cruz PI Srikanth V. Krishnamurthy, UC
Riverside Student Gentian Jakllari UC Riverside
2Premise and Motivation
- Ad Hoc Networks with omni-directional antenna
capabilities are limited in capacity - Sophisticated antenna equipment -- space-time
encoding and decoding capabilities becoming the
reality. - Antennas are becoming smaller.
- New challenge is in developing suitable network
architectures to exploit the presence of these
capabilities. - Layered approach fails to effectively exploit
these capabilities Need for layers to work
together -- a cross layered approach.
3Architectural Outlook
Discover and Maintain Appropriate Routes
Feedback To Lower Layers on Topology Construction
Network Layer
Support Scheduling based On Interference Zones
and Generated Traffic
MAC Layer
Physical Layer
Adaptive Antennas Provide Feedback To Higher
Layers On Interference Zones/ Tune In Response to
Needs
4Outline
- Objectives
- Discussion of Approach
- MAC Layer Design
- Preliminary Simulations and Results
- Routing Interactions and Higher Layer
Dependencies - Future Plans.
5Objectives
- Design and development of a cross-layer
architecture towards exploiting the physical
capabilities - Adapt to changes in traffic conditions
- Adapt to mobility
- Adapt to changes in network topology
6Approach
- MAC Layer to support searching for neighbors,
maintaining connectivity in scenarios of mobility
and flexibility to smart scheduling approaches. - Multi-path routing strategy to provide
reliability via redundancy -- tightly
intercoupled with the MAC layer. - Interactions with transport layer (or
applications) to provide adaptability to changes
in traffic patterns/requirements.
7The MAC Layer
- Full exploitation of directed (weighted)
transmissions - Many of the solutions at the MAC layer for use
with directional antennas still rely on
omni-directional reception of RTS/CTS messages. - Need to direct antennas correctly
- The communicating pair must be aware of where and
when and with what weighting co-efficients to
point their antennas correctly in order that they
can successfully exchange information. - Neighbor Discovery
- Each node should be aware of the location of its
neighbors, and the channel conditions on the link
to each neighbor -- a difficult challenge in
conditions of mobility.
8Design Approach at MAC layer
- Divide Time into three phases
- Searching -- Allow for the routing protocol to
search for new neighbors - Polling -- Periodically re-establish contact with
known neighbors to refresh channel weights and to
possibly schedule data transfers - Data Transfer --Exchange actual data with
neighbors
9The Basic Timing Diagram
Search Segment
Data
Poll
1 2
A
B
1
2
Pilot tone
Sub-slots to facilitate handshakes during search
Includes PSON / RPSON control messages to
indicate use/non-use of the particular polling
slot.
10Advantages of our Scheme
- Obviates the need for omni-directional
transmissions. - Proactive maintenance of neighborhood
information (polling) makes it robust to
mobility. - Provision of pilot tones prior to data access
allows refinement of channel weights. - Scheduled access considerably reduces the
possibility of collisions. - Provides an interface with the routing layer
(searching for and maintaining neighbors).
11Preliminary Simulations and Results
- Assume directional antennas
- Use OPNET simulator
- For now We call our MAC PMAC for polling
based MAC. - Compare with prior scheme using Circular RTS
messages Korakis et al Mobihoc 2003 and IEEE
802.11 standard. - With the Circular RTS, if a nodes position is
unknown, the RTS message transmitted
directionally and sequentially in all directions
to try and reach the neighbor.
12Simulation Parameters
Search Segment Length 20 slots
Poll Segment Length 4 slots
Data Transfer Length 800 slots
PSON 20 bytes
RPSON 14 bytes
Data Packet 512 bytes
Frame size 1.64 s
Data Rate 2 Mbps
Simulation Time 500 s
Number of Antenna Elements 8
13Simulation Results Grid Topology
- Under heavy load CRTS suffers from collisions due
to asymmetry in gain due to omni-directional
transmission of CTS messages. - In PMAC all the transmission are directional.
Thus, the gain is always symmetric and this
leads to a significant increase in network
throughput -- collisions are especially reduced
at high loads.
14Simulation Results Random Topology
- 15 nodes Nodes move with a speed of 10 m/s.
- Both protocols offer significant improvement over
802.11 - CRTS requires a high control overhead per data
packet sent (circular transmissions) while it is
much smaller for PMAC. - CRTS requires a overhead proportional to the
number of elements of the antenna array while for
PMAC it is constant. - Thus, one would expect that the benefits with
PMAC would increase with narrower beamwidths --
also applicable with adaptive antenna arrays.
15Topology Control
- Polling each and every neighbor is expensive --
too many polling slots. - Whom to poll ? -- Traffic pattern dependent.
- Input from higher layers.
- Connectivity also depends on routing -- if a
route is via a neighbor, link to the neighbor to
be maintained.
16Routing Enforcing Parallelism
- Our objective is to integrate the MAC protocol
with a multipath routing protocol. - Compute maximally disjoint paths to provide for
parallel streaming if possible. - Can reduce delays (improve quality of service)
during congested periods -- allows for bypassing
congested areas.
17Interactions with MAC layer
- Tuning the topology based on routing decisions.
- In times of partitions include an aggressive
search phase to look for new neighbors or paths
to excluded partitions. - Choose paths (and therefore links that ought to
be maintained) based on the optimization of
desired metrics of interest -- energy efficiency,
sustained data rate etc.
18Future Plans
- Inclusion of appropriate physical layer models
into simulation framework. - Include adaptive antennas (space-time processors
into framework). - Provide a unified framework with
MAC/routing/transport layers. - Investigate the interactions between layers and
tune design choices. - And of course, work with other team members
towards ensuring the success of this project !