You are here
Mobile Linux, OSS and Maemo/MeeGo Systems
The goal of the project is porting and developing of the tools for debugging of kernel and user space application for Fremantle (maemo 5). All development will be performed using beagleboard (www.beagleboard.org) device.
Three main branches in the project are:
- Porting, testing and packaging of Systemtap (http://sourceware.org/systemtap/).
This utility uses instrumentation for gathering information for a live running linux kernel. Initially it was developed for x86. So some effort should be spent to make it usable on maemo.
- Development of GUI for endurance testing.
The method of endurance testing was used by our team for finding leakages in applications. More detailed description of this project could be found here (http://www.fruct.org/index.php?morus_itemid=39).
During project it became obvious that current set of endurance tools could be used only by experienced user. That is why it was decided to develop graphical frontend for the tools from sp-endurance and sp-endurance-postproc packages to simplify testing procedure.
- Development of tools for finding resource leakages on maemo.
Endurance testing could only show that resources are leaked, but does not give any idea, where in the sources buggy code lies. There is a set of tools, that could help. They work in the following way.
Tool connects to executing process and dumps the stack of function calls, when tracked event occured. Such event could be, for example, "malloc". All tracked events and appropriate backtraces then dumped to the log. This log could be post-processed for finding resource leakages and gathering different statistic of resource usage (e.g. duration of lifetime of some resource, what function consumes more memory than others). These tools are: functracer, libleaks, mpatrol and latrace. Each tool has its own set of features and issues. Our goal is to extend and patch some of these tools for tracking of different kinds of resources like memory, gobjects, files etc. Also we are developing toolchain, where logs from different tools could be converted to unified format, post-processed and visualized.
The project team includes:
Nowadays mobile and ultra mobile devices able to maintain a wireless connection with each other and to form a client mesh network are becoming more and more spread. The basic problem lies in optimal bandwidth sharing between a management traffic and a user one. On one hand it is necessary to provide effective routing and quick reaction on unit porting/disconnection, on the other hand a high-level service of payload data transfer is extremely important too. There is 802.11s standard draft. We are going to implement simulation model of several 802.11s features. Our goal is: to find and explain 802.11s standard's limits.
Only client mesh networks without any infrastructure elements such as servers, hubs, access points etc are being considered.
We'll analyze only portable battery powered devices for ad hoc mesh networking. The main investigation directions are:
- Network/connection constancy in terms of power save mode;
- Deep and light sleep possibility;
- Beacon interval value Impaction to network performance with different power save schemes;
We are going to use NS2 simulation framework (or probably, other network simulator)
January/February 2009: 802.11s standard draft analysis
February/March 2009: Giving experience witn NS2 and WLAN modules
March 2009: Improving simulator trace output
April/May 2009: Simulation Model for beacon feature
Autumn 2009: Public report on FRUCT
Winter 2009: Devices' sleep possibilities model
Spring 2010: Public report on FRUCT
We are going to create models for new features and revisions of 801.11s standard. One of long-term research directions is an adequate mesh networks model development which would take into account the particular qualities of modern wireless devices and ways of their usage (content exchange, teleconferences, games etc). In the context of model development it is intended to observe current states of algorithms, standards and routing protocols realizations for Ad Hoc and Mesh networks. The model should include:
- QoS control,
- dynamic topology reconfiguration,
- management traffic volume control,
- load balancing.
Project homepage on OSLL site is http://osll.spb.ru/projects/show/ns2
WebDAV is universal protocol intended for providing file access over HTTP protocol. Its usage looks promising for organizing mobile devices collaboration and file sharing between them. We get flexible and extensible replacement for base file transfer protocols (like bluetooth file transfer service). Computing power and OS functionality of modern mobile devices allow us to run lightweigh network servers on them.
Accordingly there are wide future trends for creating infrastructure, that supports collaboration and file access, based on mobile devices combined into dynamic networks. These abilities may find a use for spontaneous meetings holding, mobile teams workflows organizing and just for persons, who want to keep some set of documents published.
Key aspect of these scenarios is simplicity of files publishing and getting access to them. The former shall be provided by the development of simple and user friendly utilities for server configuration. The latter can be provided on the one hand by protocol itself (as access is possible througth any web-browser) and on the other hand by the development of special utilities for remote file operating.
- Creating of a lightweight WebDAV server or its porting to Maemo platform.
- Creating of a simple, user-friendly GUI for server customizing and setting file sharing options.
- Creating of a simple, user-friendly WebDAV client for Maemo.
- Appraisal of demand on file sharing software facilities for mobile platforms.
- Development of employment methods and usage scenarios of such facilities.
- Development of software prototypes that implement former methods and scenarios.
- Review of existing WebDAV-servers and client-side libraries. November, 2008
- WebDAV server porting to Maemo platform. December, 2008
- GUI development for server customizing. January-February, 2009
- Maemo WebDAV client development. February-May, 2009
- Prototype creation. April, 2009
- Developed software publication(on Maemo garage). Making decision about further project advancing. Summer, 2009
- Preparing report and appear on FRUCT conference. Autumn, 2009
The goal of the project is to provide WidSets widgets to the Maemo platform and to make them integrate nicely into Hildon user interface. The two key problems of the project are creating performance-effective port of WidSets software on Maemo and developing a software layer to integrate widgets into the tablet desktop environment.
The Maemo port of WidSets software will use GCJ Java-to-native code compiler, GNU Classpath (Java API implementation), and MidPath library (MIDP2 implementation). Modifications will be made to these open-source libraries. The modifications are necessary to maintain compatibility with WidSets software and to provide integration with the Maemo platform.
Timing and Deliverables
02.09. Meeting with instructor, project planning, analysis of WidSets porting and integration possibilities. Requirement modeling and specification.
03.09. Design the project architecture.
04.09. Coding, testing.
05.09. Testing, debugging. Attestation. Final presentation.
Summer '09 (optionally) Experimenting.
Autumn '09 (optionally) Publishing the code in the MAEMO garage. Project report paper for FRUCT seminar. Decision on continuation of the project.
Instructor: Mikhail Kryshen
- Stanislav Epifanov, 3rd year student, responsible developer;
- Kirill Ivashov, 3th year student, developer;
- Alexander Sannikov, 3rd year student, developer;
- Dmitry Sidorov, 3rd year student, developer, project secretary;
- Lev Terentyevf, 1st year student, developer.
Nokia's N8x0 has 2.6.21 kernel and wifi driver (without mesh support), but there are opensource mac80211-based driver (published at 8.09.2008 by Nokia). As you know, since 2.6.26 kernel linux contains some mac80211-code for support mesh connections.
We are going to make Nokia N810 internet tablet as 802.11s mesh capable devices using open-source implementation of 802.11s (open80211s). The main target is to get the first practical experiences of the possibilities of the technology and services it allows.
1. Kernel update
We are building and installing new kernel on N8x0 and setup mesh. Target date: December 2008
We are testing communication between several N8x0 devices by using driver debug interfaces and/or wireless monitoring. We will try to make testing ramework for WLAN investigation. (Public report on FRUCT). Target date: March/April 2009
We are testing features of open-source implementation of 802.11s and detect areas for improvement.
- Testing heterogeneous mesh networks
- Improvement of routing procedures
- Creating full-functional testing tool-set (framework).
- Connect implementation activities with simulation modeling in NS2 (Project: )
Project homepage on OSLL site is http://osll.spb.ru/projects/show/n8x0-mesh
Current project status (updated 02/15/2009) putted on project wiki page http://osll.spb.ru/wiki/n8x0-mesh
OSLL is a group of students, postgraduates, professors and independent professionals in Computer science and Software development. Initially, it has been grown on Computer Science Department of Saint-Petersburg Electrotechnical University. Current OSLL activities are concerned with wireless networks and network simulation, knowledge engineering based on ontologies, serverless network file system, OLPC/XO, Nokia MAEMO internet tablet, gcc vectorization.
- OSLL hasn't centralized web resource for project presentation;
- All information, documentation stored in non-reliable personal pages and home compurters;
- New students (potential participants) hasn't access to information;
- Make a web resource for collaboration and development process
- Present all information for our activities and take access to all our results and sources
- Support reliable RCS storage for sources
- Support communication abilities (forums, mail-lists and so on)
- To involve more and more students and developers for OSS development in Saint-Petersburg
1st of December runs on the website osll.spb.ru
Future: support and content management
Project leader is Paul Zubarev
Maemo is a Debian-based development platform for handheld devices (e.g. N770, N800, N810). It is built from free and open source components, which are well-known in desktop Linux world. Such ideology simplifies application porting.
The goal of the project is long-term fault testing of the Maemo based Nokia device (most probably N810). Long-term testing means to run the device and periodically do its status snapshots. Analysis of snapshots shows the resource leakage problems, e.g. memory leaks. Test rounds should be long enough (several days or several weeks). During the test device is driven by the automatic testing script.
For performing of this goal the framework for automated testing should be developed. The framework should re-use and enhance already existed automated testing facilities created for Maemo.
- The following areas could be covered by testing:
- Video and Audio calls
- Office activity (Notes, PDF + mix of everything)
The expected results of the project could be classified in two categories: practical and theoretical.
The practical achievements should include:
- a number of automated tests which can be re-used by other testers
- approximate time of stabilized period (rare failures) duration
- weak areas, hardware or software (e.g multimedia or something else), also taking into account visibility for end-user
- low level information about faults (logs), minimally processed statistics
The possible theoretical findings are optional and could include ideas in the following areas:
- testing approaches
- ault detection, recovery and hiding methods
- fault injection and propagation for robustness testing
Graduated: 5th FRUCT seminar