Ch.12 Communication Systems

1
Ch.12 Communication Systems

• Valvanos text

2
12.1 Introduction

• SCI can be used to create a network.
• Figure 12.1 gives a general visualization.
• At lowest level frames are transferred between
  I/O ports, along the physical link.
• At the next level the operating system of one
  computer sends a message or packet to the OS of
  another computer.
• The message protocol will specify types and
  formats of the messages.

3
12. 1 (cont.)

• Messages typically have
• Addresses
• Synchronization or handshake field
• Data field
• Error detection and correction field

4
12.1 (cont.)

• Highest levelcommunications between users, or
  high level software tasks.

5
12.2 Reentrant Programming and Critical Sections

• If two threads access the same global memory, and
  one is a write, then there is a causal dependency
  between the threads.
• Shared global variables are important because
  they are required to pass data between threads.
• Hard to find bugs.

6
12.2 (cont.)

• A program segment is reentrant if it can be
  concurrently executed by two or more threads.
• Implementation of reentrant program segments
  avoids storing in global memory.
• Also, variables are placed in registers or on the
  stack.

7
12.2 (cont.)

• A nonreentrant subroutine will have a section of
  code called a vulnerable window or critical
  section.
• Errors
• (1)One thread calls the non-reentrant subroutine
• (2) It is executing in the critical section when
  interrupted by a second thread.
• (3) The second thread calls the same subroutine.

8
12.2 (cont.)

• Possibilities
• The second thread is allowed to complete the
  execution of the subroutine, then the first
  thread finishes the subroutine.
• The second thread executes part of the
  subroutine, is interrupted, and then re-entered
  by a third thread the third thread finishes, the
  second thread finishes, then control is passed to
  the first thread and it finishes.

9
12.2 (cont.)

• Program 12.1 illustrates a non-reentrant
  function, since num is a global variable used by
  the functions (in assembly and C).
• (In C) short num
• short Ave (short first, short
  second)
• num first
• return (numsecond)/2
• This is non-reentrant because of the access to a
  globalnum.

10
12.3 Interthread Communication and Synchronization

• A simple inter-thread scheme is a mailbox.
• Figure 12.2 illustrates an input device
  interfaced using interrupt synchronization.

11
12.3.4 Producer Consumer Problem

• First In First Out (FIFO) circular Queue and
  double buffer.
• Data Structures that can be used to link a
  source process (producer) with a sink process
  (consumer).
• Fig. 12.4 (page 440)
• Table 12.1 (examples of producers/consumers)

12
12.3.4 (cont.)

• A first in first out circular queue (FIFO) is
  useful for the implementation of a buffered I/O
  interface.
• After initialization, the queue puts and gets
  data.
• This allows the foreground thread to run while a
  second thread runs in the background.

13
12.3.4

• A data flow graph is shown in Figure 12.5.
• There is buffered input and buffered output
  illustrated.
• Valvanos textFIFOs are implemented using
  statically allocated global structures.
• An advantage of the FIFO is that the producer and
  consumer threads can be decoupled this will
  improve system performance.
• Without the FIFO, data will have to be processed
  one at a time.

14
Timing

• Let tp be the time (in sec.) between calls to
  Put, and rp be the arrival rate (in bytes/sec)
  into the system.
• Let tg be the time (in sec.) between calls to Get
  and rg be the service rate (in bytes/sec) out ot
  the system.

15
Timing (cont.)

• Then rg 1/tg and rp 1/tp.
• If min tp max tg, then a FIFO is not needed and
  a mailbox can be used.
• If the average input rate is larger than the
  average ouput rate then a FIFO will eventually
  overflow, no matter how large the FIFO is.
• But if rp is temporarily high or rg is
  temporarily low, (and the FIFO becomes full) then
  this problem can be solved by increasing the FIFO
  size.

16
Checkpoint 12.5

• If the FIFO becomes full, can the situation be
  solved by increasing the size?
• Answer
• No, if the average producer rate exceeds the
  average consumer rate.
• Yes, if the temporary producer rate exceeds the
  temporary consumer rate.

17
Figure 12.6

• This shows FIFO queues that are used to pass data
  between threads.
• OutChar is called by the main program when it
  wishes to output data.
• InChar is called by the main program when it
  wishes to input data.

18
12.3.4 FIFO Queue Implementation

• Figure 12.7 FIFO implementation with infinite
  memory.
• Program 12.3 illustrates some code fragments.
• Figures 12.8,12.9 showing Put and Get operations
  with a pointer wrap.

19
12.3.4Double Buffer

• Figure 12.11 shows the storing of data in one
  buffer while data is read from a second buffer
  and processed.
• A double buffer is two buffers of fixed size.
• Example Disk (see figure.)
• Disk has fixed size blocksbuffers are the size
  as the blocks.

20
12.4 Serial Port Interface Using Interrupt
Synchronization

• Example connect to a PC serial port.
• Baud rate must match.
• FIFOs are used.
• Initially just the receiver is armed.
• Transmitter will be armed when data is available
  within the SCI_OutChar routine.
• Interrupt occurs when new data arrives in the
  receiver data register, and an interrupt occurs
  when the transmit data register is empty.

21
Program 12.5

• Page 448, implementations of an interrupting SCI
  interface.

22
12.5 Distributed Systems

• Three simple communication systems using the SCI
  port.
• If distances are short, half-duplex can be
  implemented with simple open collector or
  open-drain TTL-level logic.

23
System 1

• Master-slave configuration
• Master transmit output is connected to all slave
  receive inputs (Fig. 12.12, page 449)
• Master broadcasts commands.
• Slaves can respond, one at a time.
• When Master transmits, it is to all slaves.
• If slaves only transmit after being triggered by
  the master, there will be no collisions.

24
System 2 Ring Network

• Transmitter and Receiver lines are connected in a
  circle.
• Figure 12.13three microcomputers connected using
  the SCI in each.
• Simple to build but slow performance (response
  time and bandwidth).
• It is also difficult to add/subtract nodes.

25
System 3 Multi-drop

• Figure 12.14, page 451
• Software activates the SP483 Driver and outputs
  the frame.
• Half-duplexthe frame can be echoed back.
• Errors can also be discovered using parity checks
  or checksums.

26
Checkpoint 12.11

• How can the transmitter detect a collision had
  corrupted its output?
• ANSWER The frame sent by a transmitter is
  echoed to its own receiver. If the data does not
  match, or if thre are any framing or noise errors
  then a collision has occurred.

27
Checkpoint 12.12

• How can the receiver detect a collision had
  corrupted its input?
• ANSWER Parity could be used to detect
  collisions. Also the message could have a
  checksum added. Framing or noise errors can
  also indicate a collision.

28
12.6 Design an Implementation of a Controller
Area Network (CAN)

• 12.6.1 CAN
• A serial data communications bus
• Used for real-time applications.
• Operate at speeds up to 1 Mbps.
• Error detection capabilities.
• Developed by Robert Bosch for use in autos.
• Now used in industrial automation and control
  applications.

29
12.6 CAN (cont.)

• Has been standardizedISO 11989.
• Figure 12.15CAN base system using 9S12s.
• Can have up to 112 nodes.
• Four Components
• (1) Bus2-wiresCANH and CANL termination is a
  120ohm resistor.
• (2) Transceivercan transmit and receive on the
  same channel.
• (3) CAN Controller
• (4) Software used to transmit and process data.

30
12.7 Integrated Circuit (I2C) Interface

• Historically, users want microcontrollers to
  require less power and space for embedded
  systems application.
• Two Ways
• Integrate functionality into the microcontroller.
• Reduce the number of I/O pins.
• I2Cproposed by Phillips in the 1980s
  originally use to interface devices using
  2-wires.
• 1998 v1 became an industry standard.

31
12.7 (cont.)

• Figure 12.19shows network made of masters and
  slaves.
• Both lines are bidirectional.
• SCLserial clock line.
• SDAserial data line.

32
12.8 Wireless Communications

• Design concerns
• Bandwidth
• Distance
• Topology
• Security
• Figure 12.26
• Short Range Device
• SPI interface is illustrated

33
(No Transcript)