Professor Izhak Rubin

1
Unmanned-Vehicle Aided Multi-Tier Autonomous
Intelligent Wireless NetworksMobile Backbone
Networks

• Professor Izhak Rubin
• Electrical Engineering Department
• UCLA
• August 2005
• rubin_at_ee.ucla.edu

2
FORCEnet Architecture using AINS Technologies

• Development of AINS system architecture for
  realizing FORCEnet using intelligent autonomous
  collaborating agents embedded in entities that
  perform communications networking, sensing,
  maneuvering and striking functions.

3
AINS Innovative Networking Technologies enable a
Network-Centric C4ISR Operation

• Development of survivable and autonomously
  adaptable mobile communications network systems
  that support high quality transport of critical
  messaging flows and real-time streams in an
  adverse environment to enable network centric
  combat operations and warfare.

4
Our Approach

• Breakthrough methods to guide intelligent
  platforms to rapidly mitigate network system
  gaps, substantially re-constitute degraded
  configurations and enhance performance, at the
  right place at the right time.
• Such methods include the autonomous layout and
  control of unmanned networked platform formations
  and UAV swarms in a multi-tier hierarchical
  mobile backbone networked infrastructure, and the
  formation of internets-in-the-sky.

5
Our Innovative Networking Technologies I

• UV aided Mobile Backbone Networks (MBNs)
  Multi-tier adaptive autonomous networking
• Robust survivable QoS Routing for mobile ad hoc
  wireless networks employing multi tier UV swarms
• Architecture, infrastructure and approaches for
  the configuration of UAV platforms and swarms to
  jointly best support
• Communications networking
• Sensing tasks
• Area search and surveillance

6
Our Innovative Networking Technologies II

• Power-control spatial-reuse Medium Access Control
  (MAC) protocols and algorithms
• Integrated MAC scheduling, power control and
  routing leading to significant enhancements in
  the throughput efficiency of shared radio
  channels
• Integrated System Management (ISM)
• New paradigm in the design of system management
  architecture that combines monitoring, control
  and resource allocations for C4ISR systems

7
Robust Wireless Networking Architecture and
topology Synthesis

• Synthesis of a multi-tier (land, air and sea
  based) mobile backbone network (MBN)
• New distributed algorithms to configure the multi
  tier backbone network
• Dynamical adaptivity to failures, application
  mixes and capacity requirements

8
Hierarchical Configuration of UV-aided Mobile
Backbone Network (UV-MBN)
ANet 1
Backbone Node Gateway
ANet 2
9
AINS based UV-aided Dynamically Reconfigurable
Network

• mbns.exe

• mbns.exe

• UV aided Mobile Backbone Network Protocol
• (MBNP)
• Quality of Service (QoS) UV-aided operation
• MBN based On Demand Routing with Flow Control
  (MBNR-FC)
• Swarm Networking

10
Illustration of our heterogeneous Mobile Backbone
Network (MBN)
11
UV aided Autonomous Mobile Backbone Network
12
Backbone Construction
                                     
                                     
                                     
                                     
                                     
                                     
                                     
                                     
13
The MBN Topology Synthesis Algorithm (TSA)

• Neighbor Discovery
• Every node exchange Hello Message periodically.
  Short timer
• Every node updates its neighbor list
  periodically. Long timer
• Each node learns its 1-hop neighbor information
  and 2-hop BN neighbor information.
• Association Algorithm
• Every node that is in a BCN state or RN state
  attempts to associate with a BN with highest
  Weight.
• The Weight of a node can be based on its ID,
  degree, congestion level, and a nodal/link
  stability measure.
• If no acceptable neighboring BN is detected, try
  BCNs If no BCN either, try RNs

(BCN 3,6)
Hello
Hello
(BCN 1,3)
(BCN 4,5,7)
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
(BCN 1,2,5,7)
Hello
Hello
Hello
Hello
Hello
(BCN 3,4,6)
Hello
Hello
(BCN 2,3)
(BCN 6,7)
Hello Message ID, Weight, BN Neighbor List
14
The MBN Topology Synthesis Algorithm (TSA)

• BCN to BN Conversion Algorithm
• (1) Client coverage a BCN that receives an
  association request from a BCN or RN, converts
  itself to a BN.
• (2) Connectivity of the BNet A BCN node finds
  that by converting itself to a BN it will upgrade
  the Bnet connectivity.
• BN to BCN Conversion Algorithm
• (1) All of its BN neighbors have at least one
  common BN neighbor whose weight is higher than
  the weight of the underlying BN that is
  considering to convert.
• (2) Each of its BCN clients have at least one
  other BN neighbor.

(BCN 3,6)
(BCN 1,3)
(BCN 4,5,7)
(BCN 1,2,5,7)
(BCN 3,4,6)
(BCN 2,3)
(BCN 6,7)
15
MBN Topology Synthesis Algorithm Convergence Time

• The MBN topology synthesis algorithm convergence
  in constant time, of the order of O(1).

16
Total number of backbone nodes (BNs) in the
network

• The backbone network (Bnet) size is independent
  of the number of nodes in the network or the
  nodal density.
• The backbone network (Bnet) size is only
  proportional to the area size.

17
Control Message Overhead of TSA

• The control message overhead of TSA is
  independent of the number of nodes in the network
  or the nodal density.

18
Data Delivery Radio of 25 UDP flows
19
Average End-to-end Delay Performance
20
Average Data Path Length
21
Average Path Length

• We expect the employment of the MBNR scheme to
  yield a longer average path length value than
  that obtained under AODV (since routes are now
  established only across the backbone network).
  Interestingly, our simulation results indicate
  that the MBNR protocol does not always produce
  longer path lengths.
• RREQ packets are transmitted as broadcast
  packets, when such a packet experiences
  collision, no MAC layer retransmission takes
  place. Consequently, if the network is already
  overwhelmed by RREQ storm, it is likely that a
  route will not be discovered in time or that a
  non-shortest path route will be selected

(a) Stationary network (b) Mobile network
22
QoS based Robust Scalable Routing (MBNR)

• MBN based Robust Routing protocols (MBNR)
• On-demand routing mechanism that uses selective
  control packet forwarding (across the MBN) to
  discover routes
• Proactive routing for route establishment in
  smaller subnets and certain Access Nets
• Unique MBN based Flow and Congestion control
  mechanism (MBNR-FC protocol) to preserve the
  quality of service (QoS) of established flows and
  to ensure that, under overloading conditions,
  only high priority flows are supported at desired
  QoS
• Unique cross physical, MAC and network layer
  algorithms and protocols to ensure that the
  realistic nature of the wireless radio
  environment is dynamically incorporated into
  communications resource allocations and routing
  operations.
• Effective use of UGV and UAV swarms to establish
  backbone routes and to distribute control packets
• Hybrid backbone and non-backbone routing and
  flow/congestion control to efficiently utilize
  resources in areas that are not covered or are
  away from the mobile backbone and its UGV and
  UAV agents

23
MBN Routing with Flow Control (MBNR-FC)Delay
Jitter Performance Comparison among Different
Protocols
24
Network Performance packet delay and delay
jitter

• Delay jitter vs. Traffic loading
• The delay jitter is reduced as traffic loading
  rate is increased (when the network is not
  saturated). Explanation route discovery produces
  a larger delay which is different from the delay
  experienced when the route is available.
• When the network is congested, more route
  discovery attempts take place.

25
Hybrid Routing Strategy

• Capacity utilization of pure MBNR-FC
• When the number of BCNs is not able to form a
  backbone to cover the whole network area,
  backbone-only paths will limit the overall
  throughput capacity of the network.
• Allowing both backbone routing and non-backbone
  routing could fully utilize the network capacity.
• Long-distance traffic vs. Short-distance traffic
• Short-distance traffic obtains shorter path
  lengths by routing through all type of nodes,
  while long-distance traffic does not.
• Long-distance traffic obtains routing overhead
  reduction by routing through backbone network,
  while short-distance traffic does not.

26
Delay-throughput performance of MBNR-FC/DA under
2-hop Anets

• The delay-throughput performance with distance
  thresholds equal to 7 hops and 9-hops demonstrate
  a significant throughput capacity gain compared
  to that with distance threshold equal to 0-hops
  (which is obtained by pure MBNR-FC).

27
Under Development Adaptive Scheme for Distance
threshold Selection

• Adaptive scheme for distance threshold selection
• Execute in a distributed manner.
• Adjust the distance threshold according to the
  current traffic distribution.
• Procedures
• Each BN collects the congestion information of
  its own Anet the number of clients that are not
  eligible for participating in the route discovery
  process (i.e. if they or their neighbors are
  congested.)
• BNs that are within 2 hops from each other
  exchange their Anet congestion indices.
• The obtained congestion information is used by
  each BN to compute a distance threshold dth which
  it broadcasts to its Anet clients

28
High Capacity QoS MAC

• Power-control spatial-reuse Medium Access Control
  (MAC) protocols and algorithms
• Integrated MAC scheduling, power control and
  routing leading to significant enhancements in
  the throughput efficiency of shared radio
  channels
• Provision of quality of service (QoS) by
  prioritized scheduling and cross layer
  MAC/Networking operations

29
MAC Mechanisms

• Power control spatial reuse (PCSR) Medium Access
  Control (MAC) layer operations
• Scheduling based QoS based MAC mechanisms (such
  as PCSR demand assigned TDMA / FDMA / CDMA)
• Random access based PCSR techniques providing
  enhanced performance
• Directional and omnidirectional operations
• PHY-MIMO driven power control MAC operations
• Autonomous power control MAC operations using UAV
  swarms

30
Power Control Spatial-Reuse MAC DA/TDMA
large increase in spatial reuse factor
31
Throughput Analysis of our Power Control
Scheduling Algorithm (PCSA) and alternative
scheme (TPA) (for an illustrative network with 10
active nodes)
32
Uniform Traffic
10001000m area, 100 nodes, 30 flows, Fixed
Routing
In this experiment, we fix the routing in
advance so we can focus on understanding purely
the characteristics of the 802.11 MAC. DPC
offers a significantly better Throughput-delay
characteristics compared to low power
transmissions (blue) and regular 802.11 with no
power control (green).
33
Localized Traffic
400400m area, 100 nodes, 15 flows, Fixed Routing
Benefits of our distributed power control
algorithm are especially apparent when
traffic patterns are localized.
34
Cross Layer Power Control based Topology Synthesis

• What is the optimal number of APs needed for best
  network performance (in terms of throughput,
  delay, delay-jitter, packet loss ratio)?
• APs should not only be deployed to provide
  coverage but also to accommodate different
  capacity needs of nodes
• What is the optimal power to operate at?
• When is it useful to employ Cell Splitting and
  get new APs or Soft APs (a laptop configured to
  work as an AP) into the network?

35
Adaptation of AP / BN selection to the traffic
profile
When using power control the number of APs
deployed Should depend on the Traffic
characteristics in the Network. When the
traffic is mostly Long distance, its better
to Employ a fewer number of APs, and vice versa.
36
On going developments Simulation Results for
Hybrid TDMA/CSMA
Experiment with three APs, 9 flows, 3 of which
are inter-AP flows. Case 1 Hybrid scheme Case
2 Regular 802.11 We can clearly see that the
hybrid scheme delivers significant throughput and
delay benefits over the regular, non power
controlled IEEE802.11 Note inter-AP flows can
traverse paths that are as long as 3 hops
37
Integrated System Management (ISM)

• New paradigm in the design of system management
  architecture that combines monitoring, control
  and resource allocations for C4ISR systems
• Hierarchical Integrated System Management and
  control architecture using nodal, subnetwork and
  system wide monitors and control elements
• Monitoring attributes and Management Information
  Bases (MIBs) for communications, sensing, UV,
  maneuverable and strike segments
• ISM algorithms for joint resource, performance,
  failure and topology management of MBN based
  C4ISR systems using UAV swarms

38
Integrated System Management system configuration
39
Integrated System ManagementIllustration of ISM
display of status of communications, sensing and
UAV networked systems
40
ISM Topology Display
41
ISM Traffic Graph Display
42
On-Going Planned Research Works

• Power control spatial reuse MACs
• Hybrid MAC for meshed architectures
• Topology Synthesis of the Backbone Networks
• Characterization and tuning of the algorithms
  performance features and comparisons stability
  and efficiency adaptations
• MBN based QoS Routing
• Development and analysis of the hybrid MBNR-FC/DA
  scheme

43
outstanding research works

• UAV and UGV aided networking
• UAV swarms
• Cross Layer networking
• Distributed cross-layer PCSR MACs
• Integrated power control MACs and MBN based QoS
  routing
• Phy / MAC / Link / Network and topology synthesis
  cross layer protocols and algorithms
• Performance analyses and simulations under a
  multitude of multimedia applications and C4ISR
  scenarios
• Incorporation of QoS oriented network management
  schemes
• Energy aware MBN based networking