Carambola2 OpenWRT on Mac OS X Mavericks

There are some sources about building developing OpenWRT on Mac OS X, you have follow different instructions in different web sites. Here is the complete steps to build OpenWRT on Mac OS X 10.9 Mavericks. I tested it both Carambola 2 native and OpenWRT x86 architectures.

First, you need a disk image on your mac. Go to Disk Utiliy->New Image-> Case sensitive, sparse disk Image. I allocate 15GB to work on 2 different builds.

Second, you have to install XCode from App Store, and Command Line Tools or run this on terminal;

xcode-select --install

Third, install macports and install required packages by macports using commandline;

sudo port selfupdate

sudo port install gcc47

sudo port install apple-gcc42

sudo port install coreutils asciidoc bzip2 fastjar flex getopt gtk2 intltool jikes zlib openssl p5-extutils-makemaker rsync ruby sdcc unzip gettext libxslt bison gawk autoconf wget gmake ncurses e2fsprogs ossp-uuid gawk findutils gnutar

Now, everything is ready to get the source code and build.  Mount the disk we setup in the first step and cd into and get OpenWRT stable version.

cd /Volumes/OpenWRT

git clone git://

or you can clone Carambola 2 sources from github

git clone

Select gcc 4.2 to compile tools and gcc 4.7 for the rest.

sudo port select –set gcc apple-gcc42

export PATH=/opt/local/libexec/gnubin:/opt/local/bin:${PATH}

make menuconfig

If you are building carambola 2 don’t change architecture. For OpenWRT  you  can choose x86 or whatever you want. For development purposes, following options need to be enabled.

  • Advanced cofiguration options and Toolcahin options.
  • Build the OpenWrt SDK
  • Build the OpenWrt based Toolchain

Save changes and run;

make tools/install

After everything compiles successfully, select gcc 4.7 and run make;

sudo port select –set gcc mp-gcc47




How to set EDCA/DCF parameters in NS3?

Having throughput problems in OMNET++, we have to switch to the NS3 as network simulation platform. It  doesn’t have a builtin IDE like OMNET and simulation language is either C++ or python. I prefer C++.

After a long time period of  searching, I’ve managed to set DCF and EDCA parameters (like contention window) for wireless simulations (802.11 networks).  The problem is when you get Dca pointer and try to set CwMin and CwMax values, it doesn’t apply to the all nodes in the network, it has to be setup every device in the network.  For eaxample;

Ptr<NetDevice> dev = node->GetDevice(0);

Ptr<WifiNetDevice> wifi_dev = DynamicCast<WifiNetDevice>(dev);

Ptr<WifiMac> mac = wifi_dev->GetMac();

PointerValue ptr;

mac->GetAttribute(“DcaTxop”, ptr);

Ptr<DcaTxop> dca = ptr.Get<DcaTxop>();





Ptr<EdcaTxopN> edca;

mac->GetAttribute(“VO_EdcaTxopN”, ptr);

edca = ptr.Get<EdcaTxopN>();





mac->GetAttribute(“VI_EdcaTxopN”, ptr);

edca = ptr.Get<EdcaTxopN>();





mac->GetAttribute(“BE_EdcaTxopN”, ptr);

edca = ptr.Get<EdcaTxopN>();





mac->GetAttribute(“BK_EdcaTxopN”, ptr);

edca = ptr.Get<EdcaTxopN>();





OMNET++ INET 802.11e Throughput Problem

It seems Omnet++ Inet still has problems with 802.11 mac layer  implementation.  The problem I face is, throughput decreases to half when the Max Contention Window  parameter is set to same value with Min Contention Window. I made the experiment with throughput example supplied in Inet. Simulation environment is set up as 1 AP and 1 Station when cwMinData and cwMaxData is set same value throughput is measured around 2.7 Mbps. Interesting thing is it is same regardless of the value I set (3, 7, 32 …). With 1 AP and 1 Station there shouldn’t be a collision and throughput shouldn’t decrease. I asked forums and reported as bug but still couldn’t find a solution. I guess I’ll be looking for alternative simulation environment, may be ns3?

Fixing OMNET++ on Mac OS X 10.9 Mavericks

Upgrading to Mac OS X 10.9 Mavericks breaks OMNET++, some actions required to get it working correctly. Here is the steps that I followed;

  1. Install latest Xcode 5.0.1 from Mac App Store
  2. Install command line tools  xcode-select –install
  3. Install macports Mavericks build
  4. Install gcc version 4.7 via macports
    sudo port selfupdate
    sudo port install gcc47
    sudo port select –set gcc mp-gcc47
    After this point you can compile OMNET++ and get working but, I faced problems when running simulations like missing icons/images in simulation. I think Tcl library has issues with Mavericks, in my case it has to reinstall too.
  5. sudo port install tcl
    sudo port install tcllib
  6. Now, edit configure.user in OMNET++ source directory
    TK_LIBS=”-L/opt/local/lib -ltk8.6 -ltcl8.6″
  7. Now you can follow OMNET++ standard installation instructions.


Book Review: Learning OMNeT++

Last week I read Learning OMNeT++ , the book is published by Packt Publishing which is about popular network simulation software. Book is 102 pages and it is easy to read and follow, it took a few days to go over and finish it. It is available in  pdf, epub and Kindle formats which is good.

Book covers OMNET++ IDE, network structure, models, basic and compound modules, simulation environment, evaluation of simulation results and INET framework with example simulations.

It is a good resource to start learning OMNET++.  It’d be nice if wireless related topics was included too. If you want to make quick start to OMNET++, I’d recommend reading the book: Learning OMNET++.

Learning OMNET++

Learning OMNET++



Thoughts About Pebble Watch

I had my pebble watch a few weeks ago and actively using it. In general watch is really good. However it has some issues. Some pebble owners face oily looking screen but I was lucky to get working one. Battery life is 5 days in average, changing 4 to 7 days with usage. Most of the iOS notifications appear on watch except whatsapp (most important app for me) and watch doesn’t use unicode encoding. Caller ID is a really big problem, already memorised most popular phone numbers in my address book, thanks to Pebble !!!  IMG_2991 IMG_2981

WCNC 2014

One of the most important network conferences WCNC 2014 will be held in Istanbul. Our plan is to attend conference with a paper which we have been working for a while if we finalize it. I hope I’d manage to meet important people  in wireless networks.

Macbook SSD Boost

Yesterday I replaced my Macbook’s HDD with Sandisk Extreme SSD. It feels like I renewed my mac. It is 3x faster than before. So, if you don’t have SSD which it asap, it will increase your productivity at least 3x.

Sandisk SSD



Spring Framework & Java EE Part 2

This is the second part of the my post on Spring & JEE first impressions. This part is a quick glance to development environments. Netbeans 7.2 is used for JEE and IntelliJ IDEA 12 is used for Spring Framework. IntelliJ IDEA is a general purpose IDE for Android, Java, J2ME, JEE, Spring ..etc.

Netbeans provides best integration with Java EE technology stack, it handles configuration automatically and helps developers with suggestions. Whenever an entity class is annotated Netbeans checks for persistent unit, if not available it makes suggestions for persistence unit.


Also, It has also code generators for entity classes, especially ones from database tables. It generates necessary classes and named JPA queries.



Netbeans comes with Glassfish bundle, application can be deployed to the app server through one click. Everything is handled by IDE.

If you are on Spring Framework which is generally developed by Eclipse, you have to generate maven configuration and related dependencies for spring and sub spring projects. After having Spring libraries, there are configurations need to be set up for the application. Sometimes its annoying to define datasource, transactions, beans … etc. At this point IntelliJ IDEA helps to handle configurations easily. It can detect enabled modules in applications.


Syntax highlighting is amazing, it is available in almost any files, java, xml … etc. It can suggest which actions to take against problems.


entityEntity generators work smoothly with any given datasource.


No doubt that ORM is a power tool for managing data through OOP manner however, some times specific OOP queries needed. But, writing and testing these queries a little harder when compared to SQL since, there are no ORM manager. IntelliJ IDEA helps it with built in console for executing object queries inside the IDE without deploying application. The most amazing tool in the IDE (personal opinion).


This post is about comparing IDEs than technology stack. Next parts will focus on more related to technologies like Persistence, MVC, JSF and Security on Spring and JEE.