Difference between revisions of "Private: mobile tv meeting minutes"

From NMSL
 
(5 intermediate revisions by the same user not shown)
Line 1: Line 1:
 +
 +
 +
 +
 +
 
== Oct 31st, 2008 ==
 
== Oct 31st, 2008 ==
 +
 
* '''Time''': 4:00-5:30pm
 
* '''Time''': 4:00-5:30pm
 
* '''Attendees''': Mohamed and Cheng
 
* '''Attendees''': Mohamed and Cheng

Latest revision as of 11:00, 11 May 2010



Oct 31st, 2008

  • Time: 4:00-5:30pm
  • Attendees: Mohamed and Cheng
  • Todo: The following tasks are identified to be finished before releasing the software. ppl-day estimations and assignees are given.
no. description ppl-day assignee committed revision
1 implementing transmitter thread 1.5 ~2 Cheng 1664
2 C++ implementation of PSI/SI that enables both online manipulation and offline editing 3 Cong
3 RTP/UDP/IP Encapsulation for streaming MP4 files 2 Cheng 1666
4 time slicing algorithms 2 Cheng 1668
5 integrating flute server (using a recurring circular buffer) 1 Cheng 1667
6 fixing ESG to enable reception at Nokia N96 N/A Cheng
7 MPE section packing 1 Cheng 1665
8 running experiments N/A Cheng

We will review the progress in the next meeting.

Oct 9th, 2008

  • Time: 3:00 - 4:00pm
  • Attendees: Mohamed, Yi, Cong, and Cheng
  • Done:
    1. GUi and backend scripts are done.
    2. Java based receiver, which runs on Linux, is done
    3. Timeslicer is modified to support different delta-T for different channels.
    4. Two canned test cases are designed to show the energy saving and switching time reduction.
  • Todo::
    1. Yi and Cong will integrate their scripts by end of this Thursday.
    2. Cheng will work on the new base station design.
    3. Yi will add OMG BCAST ESG feature to WebUI.
    4. Yi will study the debugging mechanism of N96.

Oct 3rd, 2008

  • Time/Attendees: This week's meeting is cancelled as there is no issues need to be addressed. We will get together next thursday to see the webUI/scripts implemented by Yi and Cong.
  • Done:
    1. Cong did the use case analysis and installed the webui on the new base station (cs-nsl-08.cs.surrey.sfu.ca)
    2. Yi added some implementation details to the use case analysis. However, the information is not detailed sufficient. Yi will elaborate his design next week.
    3. Yi did some work on bringing up the tablet. (40% is done)
    4. We fixed several problems in the flute stack, including: (1) We used to send a file-description table (fdt) for each ESG file. However, the IPDC ESC document specifies that each flute session should have a single FDT. (2) We used to send FEC-OTI packets in FDT, but we should have sent them in EXT_FTI header (again, per the standard). With these changes, we have validated the ESG (both IPDC and OMA BCAST versions) using two DVB-H/T tuners (Divicatch and Hauppauge Nova-T stick) and two protocol analyzers (Divicatch and dvbSAM v3). In these two analyzers, users are allowed to click on a button and play the video over the DVB-H network.
    5. We fixed the TPS (transmission parameter signaling) bits in DtPlay to specify that we are an DVB-H network with timeslicing and MPE-FEC enabled. (Note: these two options are hardcoded for now). With this change, we are able to see our broadcast network using N96. Users can click on the network ("NSL Broadcast") to trigger an ESG bootstrap.
    6. We still cannot receive ESG files on N96. Per the limited logs observed on N96, the flute receivers did not receive any block and time out. Without supports from Nokia, we cannot do much beyond this point. We will wait for emails from Nokia.
    7. We set up a Linux-based receiver using the USB stick. We can receive, decode, and play the video over the DVB-H network on this PC (cs-nsl-c05).
    8. We achieved 20-meter broadcast range using N96 as the receiver.
  • Todo:
    1. Yi will finish his implementation by Thursday. This includes (1) setting up the webui system on cs-nsl-08, (2) checking in the scripts into subversion, (3) finalizing the documentation, and (4) making sure the code runs.
    2. Cong will collaborate with Yi on modifying the webui. Later this week, Cong may help Cheng on the receiver UI.
    3. Cheng will work on a Linux receiver that allows users to switch among channels. A potential issue here is whether we can switch IP streams without relaunching vlc. The basic receiver implementation will be done by Monday.
    4. Cheng will work on customizing timeslicer to support (1) P2OPT scheduling algorithm, and (2) SIMU scheduling algorithm. Cheng will also modify the Linux receiver to accommodate the SIMU algorithm. This will be done by Wednesday.

Sep 26th, 2008

  • Time: 4:00pm - 5:00pm
  • Attendees: Yi, Cong, and Cheng
  • Done:
    1. Cong has implemented the front-end UI, which enables users to configure broadcast channels. The configurations are stored as database files, which will be read by back-end scripts.
    2. Yi has implemented a perl script to generate psi/si table based on the database files. However, we haven't integrated Yi's script with other subsystems (including Cong's front-end UI and FATCAPS utilities).
    3. We have fixed the INT table in psi/si.
    4. We have done ESG bootstrap and fragment transport following the IPDC ESG standard. This is validated using Divicatch analyzer. However, we still fail to deliver SDP files.
    5. We have composed minimal ESG files for both IPDC and OMA BCAST ESGs. The IPDC version is working (except SDP file delivery). We are currently validating the OMA BCAST version.
  • Todo:
    1. Yi will start setting up the N810 tablet.
    2. Cong will complete use cases analysis by Wed.
    3. Yi will then document the low-level testbed design based on Cong's analysis by Thur.
    4. Cheng will fix the ESG.
    5. We hope to have the demo ready by our next meeting on Oct 3rd, 2008.

Sep 19th, 2008

  • Time: 3:00pm - 4:20pm
  • Attendees: Yi, Cong, and Cheng
  • Done:
    1. Yi showed us his testbed setup, and demonstrated file transfers using flute (over DVB-H).
    2. Cong has designed the web GUI design for the program table.
    3. Yi has got an initial script to generate ESG and psi/si tables based on the program table.
  • Todo:
    1. Cong will finalize the database implementation by mid of the coming week.
    2. Yi will develop a python (or perl) script, which interfaces with Cong's database and generate ESG as well as psi/si table at run time. This will be done by the coming Wednesday.
  • Risk items: The following problems need to be addressed (in this order) for playing video on N96 cell phone:
    1. INT content (which is a psi/si table) is missing.
    2. ESG bootstrap is not working yet.
    3. Transport (main) ESG is not working yet.