Real-time Wireless Sensor Networks (WSNs): Theory

1
Georgia Tech Broadband Institute
Real-time Wireless Sensor Networks (WSNs) Theory
Applications
Prof. Raghupathy Sivakumar, ECE
2
Objectives

• Design and develop algorithms, protocols, and
  systems for real-time and mission-critical
  communication in WSNs
• Three main challenges addressed timeliness,
  robustness, correctness
• Four complimentary goals
• Higher capacity
• Lower energy consumption
• Information Prioritization
• Reliable communication
• Demonstrate research through test-bed and
  real-life applications

3
Team Funds

• 11 Members
• Raghupathy Sivakumar (Fac)
• Umakishore Ramachandran (Fac)
• Faramarz Fekri (Fac)
• Mary Ann Ingram (Fac)
• Sandeep Kakamanu
• Sriram Lakshmanan
• Rajnish Kumar
• Junsuk Shin
• Badri Narayanan
• Nazanin Rahnavard
• Lakshmi Thanayankizil

• Other Sources of Funds
• Heterogeneous WSNs
• NSF ITR, NSF REU
• Securing WSNs
• Navy
• Multi-channel WSNs
• NSF NOSS
• WSNs with smart antennas
• NSF CCR, NSF NOSS
• Multi-scale unreliable sensor networks
• NSF NOSS

(650K total funding in 2006-2007)
4
Solutions

• Information Prioritization Real-time
  Prioritization algorithms Multi-sensor
  peer-to-peer fusion Increased timeliness
• Opportunistic Large Arrays Cooperative
  transmissions Simple amplify and forward
  Increased range, and battery power
• Reliable Broadcast using Fractional
  Transmissions Hybrid coding and routing
  Lightweight rate-less coding Increased
  reliability
• Component based Multi-channel allocation Single
  channel component level operation Optimal
  performance with key practical benefits
  Increased capacity

5
Applications
Medical (patient monitoring)
Security (digital surveillance)
Video (video over IP)
Wireless (channel management)
6
Real Time Sensor Networks challenges and
solutions

• Objective
• Reduces transmissions without sacrificing
  reliability
• Motivation
• An opportunistic large array (OLA) is a form of
  cooperative diversity where simple,
  single-antenna nodes transmit together in
  response to a source or another OLA
• OLA state of the art uses flooding. While MAC
  and routing overhead is eliminated for single
  messages, more energy can still be saved
• Proposed schemes
• OLA-T (OLA-Threshold) a threshold will be
  applied to the received power to determine if the
  node should repeat. Saves about 50 of the energy
  when compared to whole network flood
• OLACRA (OLA Concentric Routing Algorithm) -
  Upstream routing suitable for SNs. Saves about
  60 of the energy in comparison with whole
  network flood, while maintaining connectivity

• Objective
• Efficient data gathering from sensors in
  wireless sensor networks
• Efficient data broadcasting in multi-hop
  wireless sensor networks
• Motivation
• Gathering data from image sensors with
  correlated data
• Updating software in already deployed
  sensor/actuator networks
• Proposed schemes
• Date gathering Scheme Distributed Source Coding
  (DSC)
• Broadcasting Schemes
• No information about the network topology is
  available Collaborative Rateless Broadcast
  (CRBcast) a two-phase data dissemination
  algorithm employing rateless coding and
  probabilistic broadcasting in the first phase and
  a local recovery in the second phase
• Local information about the network topology is
  available Fractional Transmission Scheme (FTS)
  employs rateless coding to make it possible for
  the nodes to send just a fraction of data

• Three main challenges of real-time sensor
  networks
• Timeliness
• Reliability (Packet level)
• Correctness (Application level)

• Objective
• Using multiple channels for throughput
  enhancement of wireless sensor networks with low
  complexity
• Motivation
• Channel assignment in multi-channel wireless
  networks can increase achievable throughput by
  overcoming the problems of interference, varying
  channel characteristics, poor end-to-end
  characteristics
• At any given time a node with a single radio can
  operate on only one channel
• No hardware/MAC changes, no switching delay, no
  synchronization requirement, and no scheduling
  overheads
• Proposed schemes
• We introduce a new model for channel assignment
  known as Component-Based where all links in a
  connected component induced by the underlying
  flow graph operate in a single channel
• Although the component based model looks simple,
  we show that this model can have equal if not
  better performance over link and flow based
  approaches

• Objective
• Right information at the right time
• Motivation
• Moving from a centralized to a distributed
  solution
• Increase Information-Awareness
  (Cognitive-Awareness)
• Information Prioritization and Fidelity
  Adjustment
• Proposed Scheme
• ASAP Architecture
• Separate data and
• control network
• Priority-aware
• transport
• Using multi-modal
• sensing to do prioritization

• Application 1
• IPTV enhancement using WSN.

• Application 2
• Video surveillance application with
  information prioritization

• Application 3
• Wireless mesh network management

7
Aware-home User-aware IP Video Delivery