Skip to main content

Efficient Computation of Deadline-Miss Probability and Potential Pitfalls

Fortunately, I will attend the DATE 2019 conference to present my work.
The title of my work is:
Efficient Computation of Deadline-Miss Probability and Potential Pitfalls
This work is regarding the approximation of Deadline-Miss probability for soft real-time systems.

The abstract is as the following:
In soft real-time systems, rare deadline misses are often tolerable by the applications. Towards this, probabilistic arguments and analyses are applicable to the timing analyses of this class of systems, as demonstrated in many existing researches. Convolution-based analyses allow to derive tight deadline-miss probabilities, but suffer from a high time complexity. Among the analytical approaches, which result in a significantly faster runtime than the convolution-based approaches, the Chernoff bounds provide the tightest results. In this paper, we show that calculating the deadline-miss probability using Chernoff bounds can be solved by considering an equivalent convex optimization problem. This allows us to, on the one hand, decrease the runtime of the Chernoff bounds while, on the other hand, ensure a tighter approximation since a larger variable space can be searched more efficiently, i.e., using binary search techniques over a larger area instead of a sequential search over a smaller area. We evaluate this approach considering synthesized task sets. Our approach is shown to be computationally efficient for large task systems, whilst experimentally suggesting reasonable approximation quality compared to an exact analysis.

I will also have a poster for PhD forum about my dissertation.

Comments

Popular posts from this blog

RSB+RTEMS 5/6 with QEMU-SMP (ARM realview_pbx_a9_qemu as example)

Since I got a request regarding this blog  written in 2016, summarizing again the complete flow with the latest version of RTEMS could be a good idea. Prepare a suitable workspace according to the adopted operating system on your host ( https://docs.rtems.org/branches/master/user/hosts/index.html ):  sudo apt-get build-dep build-essential gcc-defaults g++ gdb git unzip pax bison flex texinfo unzip python3-dev libpython-dev libncurses5-dev zlib1g-dev Checkout RSB and build: git clone git://git.rtems.org/rtems-source-builder.git rsb change directory to rsb/rtems/ and type ../source-builder/sb-set-builder --prefix=<the path you like to store the built toolchains> <the name of bsp> For example, to use QEMU, I need toolchains for ARM, so: ../source-builder/sb-set-builder --prefix=/home/kh.chen/respository/build/. 6/rtems-arm This will take a while. Please ensure your connection is reliable. Add the built folder into your PATH. For example, you can add one line in ~/.bas...

[LEGO nxt] Install the Enhanced NXT firmware and Upload the OSEK excutable file with Ubuntu 64bits 14.04 in 2015

This tutorial is referred from Install the Enhanced NXT firmware on NXT . I think this is the critical part to capture the idea from the tutorial for Windows. I have tried many ways to conquer the problem of libnxt3.0, however, I still have some problems on the compilation of fwflash or something else. Though the tutorial for Windows is doable, my preference is to build up everything with Linux environment. Fortunately, I notice that NxTTool in the tutorial for Windows is really powerful. We can handle the firmware updating by using NxTTool in Linux as well! I also refer to this Japanese Blog , which inspire me a lot. As usual, install the required packages: sudo apt-get install libusb-dev:i386 libusb-0.1-4 subversion fpc Please note, here is the case for 64bits user that I change libusb-dev to libusb-dev:i386 comparing to the original tutorial. Instead of using libnxt to do the uploading, here I follow that JP Blog to get the latest version of bricxcc: (The url in blog i...

[Gem5] Full system Simulation with Official Kernel

I follow the youtube provided by Gem5 to get the inspiration. (http://gem5.org/Running_gem5#Full_System_.28FS.29_Mode) I download the file here: http://www.gem5.org/dist/current/arm/ with http://www.gem5.org/dist/current/arm/aarch-system-20170616.tar.xz Then Set up M5_PATH as the place I extract the files: "/home/khchen/full_system_images/" If the setting is correct, assigning the disk-image will finish the rest of setting for dtb and kernel specifications. So far I can only use this img (aarch32-ubuntu-natty-headless.img) as my disk-image to run up to the login interface of the ubuntu: build/ARM/gem5.opt configs/example/fs.py --disk-image=/home/khchen/full_system_images/disks/aarch32-ubuntu-natty-headless.img With the other imgs, for example like "aarch64-ubuntu-trusty-headless.img", I think the kernel and the dtb should be given manually, because it will automatically load something with aarch32 prefix which doesn't make sense: info: kernel ...