ns - Pattern

Download Report

Transcript ns - Pattern

Pattern Project
Area 3: Problematiche di instradamento
Emulatore di Protocolli di
Routing per reti Ad-hoc
Alessandra Giovanardi
DI – Università di Ferrara
Scenario
Campus: student informations, group studies, game activities
Research activities
Phase 1: Studies, theoretical and simulative (SAM),
to investigate energy efficient routing protocols
Phase 2: Emulation approach to investigate routing
protocols for ad hoc networks and to give preliminary
results on energy efficiencyinterface with SAM
Test Environments
Simulation
•Cheap and quick to assemble, can do large-scale tests,
tests are controlled and reproducible
•Can be very slow, its implementation may differ considerably from real one
Live tests
•It is often difficult or too expensive to create a real test environment
of any significant size
•Real environment tests also tend not to be reproducible
•May only be possible very late in development cycle
Emulation
•Involves hardware or firmware components of the systems
•Can give a controlled, reproducible environment for running live code
•Emulates any degree of "real-world badness" in a small, lab environment network
•Provides a means for simulations to interact with a live
environment
Emulator Structure
•Can check the performance of any kind of routing protocol,
in a wireless environment.
•Developed in the user space.
•Created an interface to the SAM simulator, where many
kinds of ad hoc routing protocols are implemented.
•Based on the libpcap library and on the use of RAW sockets.
•Integrated with a statistic module (performance indexes).
•Two logical modules:
send
packet generation and transmission
analyzer
packet dump and forwarding (router)
•Developed and tested on a cluster of Linux hosts with same filesystem
•Hosts connected by an Ethernet network, but the framework is general
(IEEE 802.11)
Emulator Structure con’td
Network Organization
ns
•N hosts identified by logical number n
N
•Packet generation can be activated in all hosts
in the network or in a subset of them (Na < N)
Na
•All N hosts have router functions
nd
Analyzer Module cont’d
Packet dump
no
Type 0x0020
?
yes
ns, nd, TTL, Timestamp and nlast-hop
evaluation
no
Channel test ok
?
yes
Packet forwarding
?
no
Packet reaching final
target  statistics
yes
TTL=TTL-1
Packet drop
 statistics
TTL=0
?
no
yes
Error message
+ packet drop
Next hop evaluation (Routing block)
+ packet transmission  statistics
Send Module
•Selection of Na transmitting hosts
•Generation of the frame:
-arrival time (Poisson distribution or deterministic)
-destination nd (uniform distribution)
-timestamp (collection of the generation time)
-TTL
-payload
•Evaluation of the next-hop (Routing block)
•Creation of the frame and insertion on the network (RAW sockets)
•Collection of statistics on packets effectively sent or dropped
Implementation problems
•Host synchronization (delivery time)
•Kernel and interface packet drop
•Transparent interface with SAM centralized structure
(packet conversion, routing informations management,
Temporization..)
•...
Synchronization
•To collect the delivery times, hosts must be synchronized
•NetworkTime Protocol (NTP) can gain in some cases an accuracy of
1ms, often it can cause a mismatch in the time synchronization up to
0.5s
•We decided to compute the delivery time at the host having generated
the timestamp, i.e., at the source node ns
ns
nd
Time signalling packet (ns=0)
•Time accuracy growths with the number of nodes relaying the packet
Statistics
•Each node can be involved in the collection of parameters needed to
evaluate final performance
•Many output files are created, characterized by a prefix identifying
the logical number n of the node
•Output files collect information on:
-packets sent or dropped at the source
-packets sent or dropped at each relaying node
-packets correctly received or not (channel errors) by the destination
•Performance indexes:
-average success probability Psucc (correctly delivered packet fraction)
-delivery time Tacc (end-to-end time)
-average number of hops in the path Hop
-number of isolated nodes
Conclusions
•Proposal of a general emulation framework for routing protocols
in wireless environment
•Work in progress: tests with a simple routing protocol (Dijkstra)
without mobility
•Future works:
-investigation and emulation of ad hoc network routing
protocols (DSR,AODV)
-attention to energy efficiency, network life time etc.