Private:progress-rahman

Summer 2013

• Courses:
  • None

June 24

• Prepared youTube dataset of 1000 videos with youTubeID, Number of Views which is an indication of popularity, Video Length and bitrate.
• Incorporated Poisson and Zipf distributions in OPNET.
• Incorporated user arrival according to Poisson distribution with a fixed mean (lambda) which is modifiable.
• Incorporated user video selection according to Zipf distribution with a fixed skewness and N which are modifiable.

June 13

• Completed the extension simulations and plotted the graphs for hybrid streaming. A copy of the report can be found here.
• Working on the extension that will drive towards the thesis.

May 27

• The System Architecture of OFDMA resource allocation and user association in heterogeneous network is here.
• The MM extension document is here.

May 24

• MBSFN Area attribute under eNodeB does not support "promoted" value. It shows a list of MBSFN Areas already defined in LTE Configuration along with disabled (which means no MBSFN support). Since no "promoted" value could not find any code to set MBSFN Area attribute with function op_ima_obj_attr_set.

• Based on this observation, we contacted OPNET support team whether it is possible to join and leave the MBSFN Area dynamically for an eNodeB from the list of MBSFN Area Profiles defined under LTE Configuration. They replied that the current implementation does not support this, and this is due to the "semi-static" nature as defined in the standard. According to their reply, "In general the 3GPP specifications for MSFBN/MBMS indicates that the assignment of the MBSFN areas and and the MBMS configuration in general is "semi-static" which means that only manual changes can be done to the configuration of the network by the network administrator, but no adaptive/automatic behavior is accounted for."

• But it is implementable, although not trivial. Their response on the implementation issue is:"Unfortunately, changing this behavior in the model is not trivial because :

  a) The 3GPP specifications their selves (at least the version used for the model) do not provide any guidance or defined procedures on how to achieve that. 

  b) As I mentioned, a lot of the estimation and internal settings of the MBSFN model occur when the simulation is being initialized at time zero. Trying to make     this on demand basis over the simulation is not trivial. Many aspects of the network are involved including capacity, for example you will need to define and implement the behavior to handle the capacity for GBR unicast bearers, this is because changing the MBMS area will change how much resources are available for GBR (this is just one among many aspects that need to be handled).

  So, in general, it is possible to achieve what you want, but you really need to study and understand the model so you can make your own design and   implementation of the changes that are needed. We cannot provide you such a plan as it requires quite amount of changes and design planning. "

• We are planning to add a new state containing the initialization part associated with MBSFN Area. We will bring this portion of code from attribute_definer which is responsible for the initialization based on attributes defined in the simulation. When we need to change the MBSFN Area based on our algorithm, we will update the MBSFN Area and will do the re-initialization by a state transition to our newly created state. Now, we are looking into the code for this approach.

• Conducted a literature review on "video on-demand mobile prefetch" and tried to accumulate ideas to incorporate prefetching into the hybrid streaming approach.

Spring 2013

• Courses:
  • None

April 05

• Still could not manage to resolve the limitation posed by OPNET. I am planning to use a different approach after returning from home. Hopefully that will expedite the implementation.

March 30

• I was doing my implementation with a very basic scenario where are the users are close to the eNodeB and the channel conditions are quite good. I have tested this scenario with our resource allocation algorithm with equal subband power allocation, round robin scheduling and also throughput maximization algorithm (assign sub-carriers to user with the best channel condition for the sub-carriers) assuming the utility curve becomes parallel to x-axis after certain data rate i.e. after certain data rate even if the data rate increases, the utility value remains the same. The preliminary results show that our resource allocation algorithm achieves maximum overall utility value over the other two algorithms.

• Then I have planned to do this experiment varying the number of users. If it shows consistent results, then I will move to implementation of resource allocation with adaptive power. But unfortunately, when I have spread the users in the cell, it started giving strange errors. I have fixed a few of them and still working to get rid of these. Now, I am stuck to some errors associated with HARQ. Our maintenance support with OPNET has expired as well and I am looking into the code to resolve the error.

March 23

• Implementation of our algorithm with equal power now satisfies minimum utility for all admitted users with reasonable utility function and data rates. The main difficulty I faced so far with the implementation is the limitation imposed by OPNET as I mentioned earlier. I fix one problem with the modification which gives rise to bunch of new problems in other places.
• Need to check the implementation works for some other q-selection methods like Round robin, fair queuing.
• Currently, I am verifying my implementation with simple scenario with 10 users, 1 application and a single eNodeB.

March 16

• Almost done with implementation for our algorithm with equal power. But not sure yet, because it's too complicated and need more experiments. The most complicated part is to bypass the OPNET limitation of assigning same user to multiple subbands with multiple MCS.

March 9

• Spent most of the time working with the paper submission.
• Started working on the implementation of the thesis proposal again.

March 1

• Used demand flow to design application instead of application profile to have better design and control over the applications.
• Still working on to overcome the OPNET limitation to assign the same user in multiple subbands with different MCS. Figured out that a single entry for LteT_DL_Frame_Elem_Info and Downlink Control Information (DCI) for each UE is the main problem. This restricts a UE to span multiple subbands with multiple MCSs, because even though a user can span multiple subbands, but can use only a single MCS. But we may need to assign different MCSs to these subbands based on channel condition. Still could not find a convenient way to solve this issue.

February 21

• After communicating with OPNET support, found a bug and fixed it.
• There is also some restriction in current version of OPNET, like it cannot assign the same UE in multiple subbands. But our algorithm may assign the same UE in multiple subbands. Working on to find work around of this limitation.
• The selected bits does not exactly reflect actual allocated bits, looking into this issue as well.
• Allocation of UEs through Downlink Control Information (DCI) to different subbands causing some difficulties.
• Instead of simple Physical Resource Block (PRB) OPNET has the concept of large and small resource blocks which contributes to additional complicacy.
• So, it is taking a bit time and I still could not manage to implement the algorithm without power allocation.

February 14

• Updated the system architecture.
• Updated the literature review.
• Made some progress in the implementation, but still could not produce any results.
• The detailed report can be found here.

January 31

• Read some more papers on OFDMA resource allocation.
• Located the code segments in OPNET to incorporate our algorithm and now working on the implementation of sub-carrier allocation.

January 23

• Read some papers to construct the literature review. The report can be found here.
• Working on the implementation. Previously, I modified a function that is responsible for channel-dependent user assignment. I also need to design user queue selection algorithm and may be downlink frame generator as well to modify the creation of downlink control information (DCI)s.

January 17

• Worked on the implementation of heuristic algorithm for resource allocation without adaptive power.
• Updated system architecture and other things. Details report is available here.

Fall 2012 (RA)

• Courses:
  • CMPT 771: Internet Architecture and Protocols

January 3

• Helped Saleh in his work with OPNET to record the MCS for all users in the system in scheduled time.
• Wrote the first draft for system architecture for my work and reported to Abdul Hasib.
• Looked into OPNET code and figured out how to obtain sub-band specific CQI along with wideband CQI.
• Explored how Channel Dependent Scheduling is done in LTE and now specifically looking into OPNET code how they handle this.

December 15

• Completed the course project.
• Now will focus back on my thesis.

December 8

• Prepared the Thesis plan.
• Designed the heuristic algorithm for OFDMA radio resource allocation.
• Worked on the course project.

November 24

• Helped Abdul Hasib in generating mobility results for the proposed algorithm as to complete comments.
• Created the EPS format figures.
• Measured the runtime for both the proposed greedy algorithm and the optimal solution. For measuring the runtime for the greedy algorithm, reimplemented it in Matlab to keep the measuring environment uniform for both algorithms.
• Worked with Saleh to help him to get started with OPNET for his implimentation with explanation of how can he implement his work.
• Worked on the design of heuristic algorithm for our work and reported to Adbul Hasib.

November 11

• Did group seminar presentation.
• Working on the Course project. It is taking enough time.
• Looking into the OPNET scheduling for LTE.

October 28

• Did a survey on empirical data search on utility values for different application data rate. There are two ways to obtain utility values for different application data rates: survey based data and utility functions based data. I found all the papers I went through using utility function based data.
• Working on Internet Architecture and Protocols project: Smartphone Network Energy Management- attempt to save mobile power by controlling wireless transmission.
• Looked into the computation of wideband CQI. It turns out that OPNET computes wideband CQI based on the entire channel condition.
• There are two ways to allocate power to sub-bands: 1) same power to all-bands which is a bit easier from developer's perspective and 2) different power to different sub-bands. It turns out that OPNET uses the first approach.
• Prepared for group seminar presentation.

October 13

• Could not work for a week due to severe back pain and fever.
• Started working on the next project proposed by Abdul Hasib.
• Studying radio resource allocation related to the project and also searching for corresponding code into OPNET.

September 28

• Worked on some requested issues to submit the paper.

September 21

• Completed simulations along with the Optimal SCOPT and reported to Abdul Hasib.

September 12

• Worked on the Evaluation Results generation and writing with Abdul Hasib for the paper.

Summer 2012 (TA)

• Courses: None

September 2

• Generated the evaluation results for single cell and reported to Abdul Hasib.
• Will start working on multiple cells and try to finish the results by this week.

August 18

• Implemented the proposed algorithm in OPNET for single cell.
• Working on result generation.

August 4

• Figured out and solved the problem how we can allocate MBMS subframes in consecutive frames in Common Subframe Allocation (CSA) period.
• Figured out where MBMS group membership for UE is registered and changed it with some test data.
• Figured out how to obtain wideband MCS for a particular UE with Link Adaptation in MBMS. In MBMS, MCS obtained by Link Adaptation is not used for MBMS transmission, rather it uses MCS from Bearer Configuration in MBMS.
• Figured out IP sub-module that handles multicast group JOIN and LEAVE requests and can now join/leave multicast group dynamically from code with some restrictions.
• Running various OPNET simulations and exploring code related to simulations.

July 22

• Did various OPNET simulations to identify different aspects of MBMS in OPNET.
• Explored OPNET code to gain knowledge how MBMS is implemented in OPNET.
• Looked into the frame/ subframe occupancy by MBMS load in OPNET and reported to Abdul Hasib.
• Read the power efficient Multicast-Unicast video-on-demand streaming algorithm over LTE Networks proposal and some other MBMS documents.

July 9

• Experimented with the OPNET Lab to control the range of eNodeBs with FTP and Voice applications.
• Still exploring the OPNET code specifically the resource allocation part. It is too large and complicated as it involves physical layer heavily.
• Wrote a parser to generate frames out of OPNET debugging information. It will help to visualize different frames.

June 24

• Explored utility curve for various applications as we have a plan to consider a mixture of different types of applications instead of a single type application with benefit of real time applications in mind.
• Still exploring the code and doing experimental modifications to learn different aspects. I found forum support for OPNET poor as no replies in the forum yet. I emailed my queries to support and they replied, "your organization's maintenance contract with OPNET has expired".
• Explored some OPNET labs from OPNETWORK 2010 and 2011 covering OPNET Modeler and LTE to gain some more details about OPNET and LTE in OPNET. Since, we don't have maintenance license, I requested one of my friends in Engineering school who's lab has OPNET maintenance license to download those labs for me.

June 10

• Looking into the coding details in OPNET.
• The initial plan is to implement the basic load balancing user association rule and compare the results with two cell search algorithms namely First suitable cell and Best suitable cell for which built in support is provided in OPNET. Then we will devote more concentration to the design of more robust algorithm and implement that.
• Getting a bit problem finding out eNodeB load and looking into more details in the code. Posted into OPNET forum with no replies and that's why digging into more details.

May 29

• Studied Mobility Management/Handover in Macrocell/femtocell LTE networks. The main challenges lie within hand-in (handover from macrocell to femtocell) which itself is another research field. It has issues like discovering neighboring femto BS, admission control and selection of appropriate femto cell for handover.
• Looked into Nokia's approach (they mainly discussed various parameters without discussing the inner details of their approach i.e. working methodology).
• Designing LTE network in OPNET.

May 15

• Installed OPNET and completed Basic and Advanced tutorials under Modeler Tutorials.
• Explored LTE Model Library (could not run on OPNET as we don't have LTE license and Saleh has contacted them about this)
• Looking into Mobility Management for Femtocells.

Spring 2012 (RA)

• Courses:
  • CMPT-880: Special Topics in Computer Graphics, HCI, Computer Vision, and Visualization (Medical Image Analysis)
  • CMPT-886: Special Topics in Operating Systems

May 1

• Completed the survey and reported to Abdul Hasib.
• Working with user-association and OPNET with Abdul Hasib.

April 17

• Working with Abdul Hasib on LTE-Advanced HetNet and explored some potential research fields.
• Cost effective deployment of relay/femto cells.
• User association to macro/micro cell to improve performance.
• Carrier component allocation to improve fairness, reduce interference.
• Working on to create literature review and going to explore Facility Location Optimization (MITACS) as well.

March 27

• Working with Abdul Hasib on LTE-Advanced Resource Allocation and Interference Management.

February 17

• Concluded applicability of MapReduce to BioInformatics dataset with some comments.
• Read the report of Ahmed Hamza and discussed with him to get acquainted with the task of 3D video processing primitives on cloud platforms.
• Explored open source solutions to figure out some which can be helpful for 3D video processing primitives.
• Discussed with Ahmed about the strategy in search of an appropriate solution.
• Here is the report.

January 27

• Ran some experiments on Fei Gao's thesis work to get acquainted with it.
• Exploring different techniques to speed up similarity measurement in BioInformatics data e.g. DNA/RNA sequences report.
• The two 8 level courses consuming too much time and the initial shuffle between TA/RA
• I believe things will improve from next week.
• Planning to explore a bit more on existing approaches with BioInformatics data including kernel based clustering algorithms as those are closely related with Fei Gao's work with comments from Wael Abd-Almageed, Ph.D in mind.

January 4

• Read the thesis paper of Fei Gao and generated the report.
• Planning to run some experiments on Fei Gao's work in the cluster of our Lab.

Fall 2011 (RA)

• Courses:
  • CMPT-705: Design and Analysis of Algorithms
  • CMPT-726: Machine Learning

December 29

• Discussed with Fei Gao about his research work and BioInformatics data.
• Reading the thesis paper of Fei Gao.

November 22

• Created Azure sample application using Azure storage and tested using my free azure test account.
• Explored abstracts and introductions of papers to find out some papers with unexplored materials to be read.
• Spent some time to figure out Machine Learning Project related to my research field.
• Produced the report.

November 9

• Read a couple of more papers thoroughly to cover another dimension and generated the report.
• Planning to look for any new idea that is not covered by looking into papers.
• Also planning to look into details of some development module like Merkel trees etc.

October 25

• Went through some papers (only the 1st phase reading) to decide some more papers to read and learn.
• Read a couple of more papers thoroughly and analyzed all the papers read so far.
• Created the report.

October 11

• Read several more papers and generated corresponding report.

September 23

• Read a few more papers.
• Created initial report on my reading.

September 6

• Wrote some sample programs on Azure for both VS 2008 and VS 2010
• Went through the paper "Ensuring Data Storage Security in Cloud Computing" as part of the survey