Distributed Embed Ed System

You might also like

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 27

Distributed Embedded Systems

Need for DES:

Different task are physically distributed (location wise)


Data reduction (Initial signal processing at event place)
Reduces load on single processor
Modularity is maintain
Easier to debug
Fault tolerance can be maintain
Distributed Embedded Architectures

Interconnection between Processing Elements (PE)


Communication between PE
Physically separated PE have interaction
One part of system can be used to diagnose problems in another
part
Reusability increases
Network Abstraction

Application Layer End user Interaction


Presentation Layer Data Formats
Session Layer Dialog controls
Transport Layer Connections
Network Layer End to End Service
Data Link Layer Reliability
Physical Layer Mech/Electrical spects

OSI Model
Hardware and Software Architecture
P-P Communication
Simplex
Duplex Half/Full
Buses
Use for communication
uses packets
have some payload
Arbitration Fixed priority
Fair arbitration (Round Robin-No starvation)
Network
Dataflow
Single stage/Multistage
Networks for Embedded Systems

•I2C Bus

•CAN Bus

•SHARC Link Ports

•Ethernet

•Myrinet

•Internet
I2C (Inter IC)Bus

SDL: Serial Data Line


SCL: Serial Clock line

S Address WR-0 Data S Address WR-0 Data P


RD - 1 RD - 1 (Slave)
1 7 1 7 1 7 1 7 bits 1
Physical Layer
Low cost and easy to implement
Moderate speed (upto 100 kbps for Std)(400 kbps for
extended)
Uses 2 lines SDL Serial Data Line
SCL Serial Clock Line
Every node is connected to both lines (SDL, SCL)
Multi master Bus: More than one master
Master: Initiates data
Slave: Only responds to request from master
Master is responsible to generate SCL clocks(High)
Slave can stretch low period of clock
Data Link Layer
Every device is determine by unique address
Addr-7 bits long
1 bit –data direction 0:M---S and 1:S---M
Data push programming style is used (polling method)

1 7 1 1
S Addr W/ Data P
R

0-W, 1-R
An I2C Interface in a Microcontroller
CAN (Controller Area Network)
Physical Layer
Designed for Automotive Electronics
Uses serial bit transmission
Speed 1Mbps over twisted pair max of 40 meters
Multi master Bus: supports multiple masters
Logical 1-Recessive and 0-Dominant
Synchronous Bus
Data Link Layer
uses CSMA/AMP (Arbitration on message priority)

S-0
P-0000000
Remote Transmission Request (RTR) 0-Read, 1-Write

1 12 6 0-64bits 16 2 7
S Arbitration Control DATA CRC ACK P

Identifier Data
length
Architecture of CAN Controller
SHARC link Ports

Ceramic Quad Flat Pack (CQFP) System:


Interconnection through serial ports
Peak rates 120MFlops to 480 MFlops
SHARC Link Ports
Uses Multiprocessor Communication
High Speed ports (120-480 Mflops)
6 Link Ports are available
Data Transfer are packed into 32/48 bit word
Link Ports are in Half Duplex
Token Passing between the processors
Link Ports can target the DMA transfers
Ethernet

Uses CSMA/CD
Network is Not Synchronized
Low Cost
Bus with Single Path
Uses Twisted Pair/ Co-Ax Cable

Ethernet Organization
Ethernet Packet format

Preamble Start Dest. Source Length Data Padding CRC


frame Addr Addr

Ethernet CSMA/CD Algorithm


Myrinet

Very High Speed Network (640 Mbps)


Operates in Full Duplex Mode
BW 1.28 Gbps
Use parallel processing
High performance network system
Uses packet transmission
Internet

Node A Node B
Application Application

Presentation Presentation

Session Session
Router
Transport Transport

Network Network Network

Data Link Data Link Data Link

Physical Physical Physical


Internet

IP Packet Structure

Header Service
Version Total Length
length type

Identification Flags Fragment Offset

Time to Live Protocol Header Check Sum

Source Address

Destination Address

Option and Padding

Payload Data
Uses IP
Uses Routers
Uses Network Layer
32 bits for IPV4 and 128 bits for IPV6

FTP HTTP SMTP Telnet SNMP

TCP UDP

IP

Internet Service Pack


Network Based Design

• Communication Analysis
• System Performance Analysis
• Hardware platform Design, Allocation and Scheduling
Communication Analysis

Message Delay
Tm = Tx + Tn + Tr
where Tx = Transmitter side overhead
Tn = Network transmission time
Tr = Receiver side overhead

Total delay
T = Td + Tm
Where Td = Network availability delay
Tm = message delay
Td depends on the arbitration used (fixed priority or fair (RR))
PE 1 PE 1

PE 1

PE 1 PE 1

• Single-hop communication Networks


• Multi-hop communication networks
System Performance Analysis
Simple task graph

Tp1 Tp2
nTx
P1 P2

Distributed system with multi-rate concurrency

PE 2
PE 1 P2 PE 3
P1 P3
P4
Hardware platform Design, Allocation and Scheduling
Hardware platform issues
• Number of PEs required
• Types of all PEs
• Number of networks used
• Type of networks

Efficient system design issues

• I/O - intensive system design


• Computation - intensive design system

Load Balancing is a good idea for the Design of DES


I/O - intensive system design
• Inventory the require I/O devices
• Determining the I/O to connect to network or not (local/remote
processing)
• Analyze communication time
• Allocate minimum required PE
• Design communication intensive processes

Computation - intensive system design


• Start with task with shortest deadlines (Scheduling overheads)
• Analyze critical communication time
• Allocate lower priority tasks to shared PEs where possible

You might also like