Transcript EE194: Wireless Networks Group #2: Joseph Cerra and Stuart

EE194: Wireless Networks
Group #2: Joseph Cerra and Stuart Peloquin
Dynamic Transmission Power
Control in Wireless Ad-Hoc
Networks
INTRO
•
Discussed
– Many forms of power conservation in
wireless networks
– Dynamic transmission power control as
the preferred method for this research
– JiST / SWANS as a simulation
environment
INTRO
•
Current work includes:
– Learning the JiST API and SWANS network
structure
– Becoming familiar with the Java programming
language (Stu)
– Building a working front end to visualize
simulation (Joe)
– Implementing needed hooks and data
structures
INTRO
•
Future work includes:
– Complete design specifications
– Simulate various networks
– Collect simulation results
– Compare, contrast and report simulation
results
JiST / SWANS
•
Java in Simulation Time / Scalable
Wireless Ad hoc Network Simulator
– Well planned simulation environment
– Very customizable
– Follows OSI model
– Open source: hooks can be added
anywhere for statistic gathering
JIST/SWANS
•
JiST API
– JiST is comprised completely of Java code
– Code written for JiST is also valid Java
– Contains proxiable entities that are
placeholders for different interface types
• Transport Interface can be proxied by TCP or UDP
etc…
• This is done at simulation time by a bytecode
rewriter
JiST/SWANS
•
SWANS is a layer built on JiST
– Implements many widely used protocols
– All layers are customizable/rewriteable
• Layers can also be dropped or added
• Simulations can take place between the
interface of any layers
SWANS
•
Components
– Physical – field, radio
– Link – mac
– Network – network address, network ip
– Routing
– Transport
– Application
– Common – message description
Physical
•
Field defines the world parameters
– Geometry: Area where nodes can be placed and move to
– Placement: Where each node is placed
– Mobility: Assign a mobility model to each node
• Static, Random Waypoint, Teleport etc…
– Fading, Path loss: Signal propagation loss
•
Radio Interface
– Radio Noise Model
• Field passes message to the radio
• Radio Interface determines how to analyze received
signal. Is it a message or noise or both.
Link / Network
•
MAC Interface
• How MAC protocol reacts to messages
received by either Physical layer or Network
Layer
•
Network Interface
•
•
•
•
Determines node addressing
Communicates to MAC layer
Communicates with Routing Protocol
IPv4 implementation
Routing / Transport
•
Routing protocols currently implemented
• Zone Routing Protocol
• Dynamic Source Routing Protocol
• Ad-hoc On-demand Distance Vector Routing Protocol
– Receives calls from network layer to determine
the next hop
•
Transport protocols
• TCP – Uses TCP Sockets
• UDP
Application
•
Extremely useful application layer
implementation
– Can use any current java application
• Java sockets are rewritten by the bytecode rewriter to
use SWANS implementation of sockets
– Each node can run any application including
• Web server, database, peer to peer application
• Random message generator
•
Node setup and statistics are handled by
the driver
Graphical User Interface
•
8 Java Classes Created / Modified
– 4 Created
• MyCanvas, MyButton, MyNode, MyMenu
– 4 Modified
• Heartbeat, RadioInfo, Field, AppHeartbeat
•
Heartbeat
– Field createSim(int nodes, field length)
• mobility, fading, pathloss, radio, field
– void createNode(i, field, placement,
radioInfoShared, mapper, pl_in, pl_out)
• radio noise, mac layer, network layer, application layer
Created Classes Modified Classes
•MyCanvas
•Field
–Pointer provided to
heartbeat
–Used exclusively for
drawing nodes
•MyNode
–Stores Location, ID and
Powerlevel
•MyButton,
MyMenu
–moveRadio
–API.Proxy()
•RadioInfo
–setPower(double)
•AppHeartbeat
–set/getPower
–Computes neighbour
statistics
–Decodes messages for
transmit power
Shown Simulation
•
•
•
•
•
Wake up
Broadcast yourself
Listen to find all neighbors
Compute # of neighbors for the last 5
awake cycles
Base power on number of neighbors
Current Work
•
•
RSSI and Transmit strength not
directly implemented in SWANS
This functionality is added through
extending the message, radio
interface and MAC interface
Current: Transmit strength
•
•
•
Transmit strength is added to the
message payload from the MAC layer
Radio Interface reads this and passes
the message along at that signal
strength
MAC layer of next hop reads and
strips this information before passing
the message along
Current: RSSI
•
Two implementations adopted for RSSI
– Message passing
– Scratch Pad
•
•
Both are implemented in current devices
Message Passing: (method of choice)
– Radio adds signal strength information to each packet
•
Scratch Pad:
– All layers have access to a pseudo layer that contains
useful information
– Includes a table of the signal strength of recent
messages
Current: RSSI - Models
Node 1
Node 2
Application
Application
Transport
RSSI
RSSI
information
passed as part
of the
message
Route
Table
MAC
Radio
Field
Transport
Route
MAC
mobility
Radio
mobility
Future: RSSI / MAC
•
•
Implement Message passing of RSSI strength
Implement neighbour table in MAC layer
– Include required power levels
•
ReqP = Tst – Rst + Pth
•
•
•
•
ReqP: Required power
Tst: Transmitted strength
Rst: Received strength
Pth: Threshold power level
– If ReqPt < ReqPt-1 Then object is getting closer, use
small Pth
– If ReqPt > ReqPt-1 Then object is getting further away,
use large Pth
Future: Simulate - Static
•
Hook a complete network together using
JiST/SWANS
• Network will be composed of a collection of identical nodes
– Each node has same transmit power
– Simulations will be run on a variety of transmission powers
•
Determine a useful application to run at each node
– Need it to announce itself and take advantage of multiple
new nodes
– Traffic pattern should be semi-uniform
•
Get power consumption statistics
• CPU active/idle/sleep time
• Radio sleep/idle/receive/send time at what power level
Future: Simulate - Dynamic
•
Modify static network with new MAC
layer
– Uses RSSI and Transmitted strength
levels to determine correct send
parameters
•
Get similar power consumption
statistics
• Expect CPU time to be greater
• Expect transmit strengths to be less
Future: Routing
•
Determine modifications to use new
information in MAC layer in the routing
protocol
•
Goal is to use current routing protocol
Link metrics for current routing protocol
updated based on required transmit
strength
•
Project Future
•
•
•
•
•
•
GUI
Standalone
Set up interfaces: Which routing? Which
transport? Which mobility?
Conform to OSI model
Include all simulation options in GUI rather
than in runtime script
Make GUI and Simulation more robust
– Add functionality, more interfaces, and more
networking components
•
Include network lifetime, CPU and Radio
power consumption
Questions?