System Programming
Final assignment
2018 - 2019
Bachelor Electronics/ICT
Last update: November 20, 2018
Academic Integrity
This is an individual assignment! It is only allowed to submit your own work. You
may discuss the assignment with others but you are not allowed to share work or use
(part of) another’s solution. If you include work (e.g. code, technical solutions,...)
from external sources (internet, books …) into your solution, you must clearly
indicated this in your solution and cite the original source. If two students present very
similar solutions, no distinction will be made between the ‘maker’ and the ‘copier’.
Programming Assignment And Code Self-Review
Sensor Monitoring System
The sensor monitoring system consists of sensor nodes measuring the room temperature, a
sensor gateway that acquires all sensor data from the sensor nodes, and an SQL database to
store all sensor data processed by the sensor gateway. A sensor node uses a private TCP
connection to transfer the sensor data to the sensor gateway. The SQL database is an SQLite
system (see lab 7). The full system is depicted below.
The sensor gateway may not assume a maximum amount of sensors at start up. In fact, the
number of sensors connecting to the sensor gateway is not constant and may change over
time.
Working with real embedded sensor nodes is not an option for this assignment. Therefore,
sensor nodes will be simulated in software (see lab 8).
Sensor Gateway
A more detailed design of the sensor gateway is depicted below. In what follows, we will
discuss the minimal requirements of both processes in more detail.
SQL conn SQLite
DB
Sensor Sensor gateway
node 2
Connection
Minimal requirements
Req 1. The sensor gateway consists of a main process and a log process. The log process is
started (with fork) as a child process of the main process.
Req 2. The main process runs three threads: the connection, the data, and the storage
manager thread. A shared data structure (see lab 9) is used for communication
between all threads. Notice that read/write/update-access to the shared data needs to
be thread-safe!
Req 3. The connection manager listens on a TCP socket for incoming connection requests
from new sensor nodes. The port number of this TCP connection is given as a
command line argument at start-up of the main process. e.g.: ./server 1234
Req 4. The connection manager captures incoming packets of sensor nodes as defined in lab
8. Next, the connection manager writes the data to the shared data structure.
Req 5. The data manager thread implements the sensor gateway intelligence as defined in
lab 6. In short, it reads sensor measurements from shared data, calculates a running
average on the temperature and uses that result to decide on ‘too hot/cold’. It doesn’t
write the running average values to the shared data – it only uses them for internal
decision taking.
Req 6. The storage manager thread reads sensor measurements from the shared data
structure and inserts them in the SQL database (see lab 7). If the connection to the
SQL database fails, the storage manager will wait a bit before trying again. The
sensor measurements will stay in shared data until the connection to the database is
working again. If the connection did not succeed after 3 attempts, the gateway will
close.
Req 7. The log process receives log-events from the main process using a FIFO called
“logFifo”. If this FIFO doesn’t exists at startup of the main or log process, then it
will be created by one of the processes. All threads of the main process can generate
log-events and write these log-events to the FIFO. This means that the FIFO is
shared by multiple threads and, hence, access to the FIFO must be thread-safe.
Req 8. A log-event contains an ASCII info message describing the type of event. For each
log-event received, the log process writes an ASCII message of the format
Connection
<sequence number> <timestamp> <log-event info message> to a new line on
a log file called “gateway.log”.
Req 9. At least the following log-events need to be supported:
1.From the connection manager:
a. A sensor node with <sensorNodeID> has opened a new connection1
b. The sensor node with <sensorNodeID> has closed the connection
2.From the data manager:
a. The sensor node with <sensorNodeID> reports it’s too cold (running avg
temperature = <value>)
b. The sensor node with <sensorNodeID> reports it’s too hot (running avg
temperature = <value>)
c. Received sensor data with invalid sensor node ID <node-ID>
3.From the storage manager:
a. Connection to SQL server established.
b.New table <name-of-table> created.
c. Connection to SQL server lost.
d. Unable to connect to SQL server.
How to start?
The sensor gateway is a kind of integration result of the code of lab 5 up to lab 9. The picture
below shows the relationship between the different components of the sensor monitoring
system and the lab sessions. It should be clear from the description of the requirements which
pieces of code of these labs are required for the sensor gateway.
1 Remark that the sensor ID is only known after the first data packet is received.
Deliverables and acceptance criteria
On https://labtools.groept.be you will find a description of what exactly and in which format
(source code files, directory structure, make file, etc.) you need to prepare and upload your
solution. Once you have finished the assignment, you upload your solution on
https://labtools.groept.be. Please remember that https://labtools.groept.be is not a
compilation nor test tool. Write test code – also non-trivial test code – yourself and
thoroughly test your code. For instance, test your solution with multiple sensor nodes running
concurrently using very small sleep times between two measurements. Include a debug-mode.
Important remark about a 1 or 2 sbuffer-solution. A 1 sbuffer solution should be your
default choice. Only when a 1 sbuffer solution is beyond your capabilities, you can choose a
simplistic 2 sbuffer solution. But since the latter makes your solution much more simple, you
will get a lower mark!
Your solution will only be accepted for grading if the following criteria are fullfilled:
Your solution is available on https://labtools.groept.be before the deadline.
Your solution includes the code self-review document in attachment. This is a
critical reflection on your own code. Be honest with yourself! The evaluation on the
exam will reveal the truth anyway.
Your solution compiles, runs and generates at least some meaningful output (e.g.
measurements in the database, some logs, …);
Your solution must be a reasonable try to implement the minimal requirements. This
excludes, for instance, solutions that consist of an (almost) empty .c file, incomplete
code, or code that has no or very little relationship to the exercise, code that
implements a different assignment (e.g. from a previous academic year), etc.;
Your source code is structured and readable. The following, for example, is not
acceptable: you don’t logically structure the code into source and header files, you
apply bad naming of variables and functions, you don’t use typedefs to create logical
names, the code is not indented well, you write ‘spaghetti-code’ with goto’s, …;
You must be able to present and defend your solution. During the defense of your
solution, you are supposed to be able to answer on the technical questions of the
evaluator. This includes technical questions related to programming (C, Linux system
calls), source code building (preprocessor, compiling, linking, Valgrind, …) and Linux
command line basics. The reference guide for the minimal knowledge on Linux are
the Linux lab manuals. You are supposed to be able to work with the commands and
tools introduced in these lab sessions. You should also be familiar with the usage of
the programming tools introduced in the labs. More concretely, we target at least the
following tools and their options:
gcc tool set (preprocessor, compiler, linker, assembly) and options
(preprocessor symbols, code optimization, debug flag, ...)
creating static and shared libraries (ar, ldd, ldconfig, gcc), linking libraries
(gcc)
valgrind
make tool and make files (you have to be able to compile and run your code
using a simple make file)
Paper Assignment
Exercise 1
Draw the memory layout of your program when the program is in the following state:
TCP connections are made on port 6543 and IP 127.0.0.1.
The room-sensor map contains room numbers 501, 602, 703, 804 and 905 that map on
sensor IDs 10, 20, 30, 40 and 50 respectively.
Main process: all threads are created and the threads are in the following state:
Connection manager: sensors 20 and 50 have made a connection and each of these
sensors have send 1 measurement (21°C).
Data manager: all received sensor data is parsed.
Storage manager: no sensor data is stored yet into the database. The storage
manager is just starting to read the first sensor data from sbuffer.
Log process: at least 1 log message is received and the process is waiting on the next
log message.
Use the template in attachment as a starting point. This template shows the general structure
for a multi-threaded process and is organized in such a way that each thread can easily access
the heap. The template also includes a few examples that show how to represent the following
variables and stack frames:
int x = 7;
struct { unsigned day, month, year; } d;
file * fp = fopen( … ); Notice that the real content where fp is pointing to is hidden.
‘printf’ stack frame: since ‘printf’ is a standard library function, you can indicate that
the frame content is hidden. You can also do this for system calls and other library
functions not implemented by yourself.
<name> or <function>: replace this with the real thread or function name
Remark: for ease of reading, the function call stack grows up in this memory layout drawing.
Exercise 2
The sbuffer data structure is shared between several threads and, as such, access to it must be
protected. Different synchronization primitives could be chosen (mutex, semaphore, barrier,
condition variable, r/w locks, …) to do this job. What did you choose? And maybe more
important: why? Write a critical reflection on your choice of synchronization primitive or
combination of primitives.
Motivate the correctness of your solution.
Is deadlock avoided?
How can you be sure that multiple writer-threads can’t access sbuffer at the same
time?
Is your synchronization solution ‘fair’? For instance, if a writer or reader thread
wants to access sbuffer, it should get it immediately if sbuffer is not locked by
other threads. Of course, the starvation-problem should be avoided.
? Argue why your choice is the most optimal one. Does your choice guarantee efficient
use of the CPU?
Make a drawing that illustrates the general CPU usage of all threads in relation to
received sensor data. Use the template below to do this. You draw a box to indicate
the process state of the thread is ‘running’ (The other process states like blocked or
waiting are not shown here.). You may not assume that the sensor data arrives with a
fixed period, as clearly indicated in the template. A few example cases are added to
illustrate the idea:
case 1: threads are executed one after the other (no concurrency);
case 2: some threads are executed concurrently;
case 3: one thread seems to apply a polling strategy;
case 4: two threads seem to block on some event;
CPU usage: case 1
Sensor data arrival time
Connmgr CPU usage
Datamgr CPU usage
Storagemgr CPU usage
CPU usage: case 2
Sensor data arrival time
Connmgr CPU usage
Datamgr CPU usage
Storagemgr CPU usage
CPU usage: case 3
Sensor data arrival time
Connmgr CPU usage
Datamgr CPU usage
Storagemgr CPU usage
Deliverables and acceptance criteria
Take the following thoughts into account when designing a solution:
– The drawings should represent the code of your solution, not some other code or some
creative idea that you have in mind but you didn’t implement.
– Be aware that your drawings become easily unreadable. Re-factoring your drawings at
the end is a necessity.
Your solution will only be accepted for grading if the following criteria are fullfilled:
A hard-copy (print on paper!) of your solution is available on the oral defense of your
assignment. The result of exercise 1 should be printed on A3 format.
The documents include a header containing your last name, first name and your
student number.
Your drawings are readable.
The result of exercise 2 should not exceed two A4-pages.
Due Date of the Assignment
You must defend your solution on the date and location stated on the examination schedule.
To create a more efficient planning, you must subscribe before the due date of this
assignment to one of the events available on <link will be included later on>.
Due date of the programming assignment for the 1st evaluation
period: January 9, 2019 before 17 p.m.
Solutions uploaded after the deadline will be rejected. The code used for the defense must be
the same code as uploaded and tested on https://labtools.groept.be. No other code will be
accepted.
CPU usage: case 4
Sensor data arrival time
Connmgr CPU usage
Datamgr CPU usage
Storagemgr CPU usage
Due date of the paper assignment: oral defense of your assignment
Calculation of the final mark
The final mark is computed as:
25% on the paper assignment
75% on the programming assignment.
Remark that a solution based on 2 sbuffers will result in a lower mark (max. 15/20) on the
programming assignment.
版权所有：编程辅导网 2021 All Rights Reserved 联系方式：QQ:99515681 微信：codinghelp 电子信箱：99515681@qq.com
免责声明：本站部分内容从网络整理而来，只供参考！如有版权问题可联系本站删除。
