[Ns-developers] GSoC 2010 - MAC and PHY models - work plan

Sat Apr 3 10:33:26 PDT 2010

Hi Giuseppe,

sorry for the late reply. Overall, your proposed plan looks good to me.
Here are some general comments:

1) it is not clear to me whether the proposed "steps" are sequential or 
you plan to develop them in parallel. Maybe the best solution is a 
mixture of the two. For example, you could do most of step 1 at the 
beginning but with no propagation model, then just do step 2 and test 
communication among the devices (e.g., to see that OFDMA works), and 
develop propagation modules only afterwards.

2) similarly, it would be good to have "milestones". E.g., after steps 1 
and 2, one milestone could be to develop a test program in which you 
perform point-to-point communication between an eNB and a single UE.

3) according to the ns-3 policy for the GSoC 
http://www.nsnam.org/wiki/index.php/GS02010StudentGuide
students will have their code reviewed twice (one in the middle and one 
at the end of the GSoC). You should give us some indication of what 
steps you expect to be completed for the first review. Of course we 
understand that this might need to be adjusted during the development of 
the project.

Regards,

Nicola

Giuseppe Piro wrote:
> Dear all,
> 
> Please, find in the following my work plan for GSoC 2010
> 
> PROJECT: GSoC 2010 - NS3 - MAC and PHY models (mentors: Nicola Bandlo 
> and Marco Miozzo).
> 
> PROJECT IDEAS: develop a framework to simulate LTE networks on NS3. The 
> framework will be composed by 4 different components:
> - a standard compliant LTE PHY layer 
> - LTE network devices (UE and eNB) 
> - basic user plane protocol stack
> - Frame Management and Downlink Resource Allocation 
> 
> 
> WORK plan:
> here there is my proposed work plan:
> 
> STEP 1: DEVELOP a standard compliant LTE PHY layer.
> 
> To this aim I should extend the Nicola's Spectrum Framework, according 
> to 3GPP specifications [1].
> 
> The FDD access mode will be implemented, considering a bandwidth equal 
> to 10 MHz @ 2 GHz (the most used in the research community). Obviously , 
> this parameter could be modified in the future, in order to simulate all 
> permitted bandwidths for LTE. 
> The whole bandwidth will be divided into two equal size sub-band for the 
> DL and the UL  respectively. 
> 
> The SpectrumModel will be modeled to obtain a set of sub-band, each one 
> lasting 180 KHz.  At each UE netDevice, the SpectrumChannel::StartRX of 
> the PHY layer will be used to obtain the SINR for each sub-bandwidth. In 
> particular several PHY models will be implemented or will be merged with 
> the already existing PHY layer to obtain this objective.
> The SINR will be matemathically obtained using the approach proposed in 
> [2]. 4 different PHY models will be considered:
> *Pathloss: PL = 128.1 + (37.6 * log10 (R)), where R is the distance 
> between the UE and the eNB in Km.
> *Multipath: Jakes model
> *PenetrationLoss: 10 or 20 dB, depending from the user speed and channel 
> model [1]
> *Shadowing: log-normal distribution (mean=0dB, standard deviaition=8dB)
> 
> The SINR will be converted to the CQI feedback (one for each 
> sub-bandwidth) using PHY TABLE, as proposed in [3]. CQIs will be sent 
> periodically to the eNB using uplink bandwidth and constant MCS. No AMC 
> module will be implemented for the uplink during the GSoC. 
> 
> For the eNB network device, the AMC module will be implemented, in ordert to
> (i) choose the proper MCS to send packets of scheduled flows at the MAC 
> layer (see after for details)
> (ii) define the quota of data (Transport Block Size) that each scheduled 
> flows should send every TTI.
> To this aim, the AMC module will use PHY TABLE as proposed in [3].
> 
>    
> STEP 2: develop LTE network devices (UE and eNB).
> 
> Create UE and eNB networkDevices extending the netDevice class. Each 
> device should 
> (i) enqueued packets from the upper layer
> (ii) forward packets to the upper layer
> (iii) send and receive packets from the PHY layer
> (iv) manage other LTE parameters, LTE functionalities etc...
> 
> 
> STEP 3: develop basic user plane protocol stack    
> 
> For each netDevice will be defined a basic implementation of the user 
> plane network protocol including:
> MAC, PDCP, RLC (TM mode at the beginning and UM mode if there is 
> available time at the end of GSoC). 
> For the GSoC, HARQ will not be developed. 
> 
> 
> STEP 4: Frame Management and Downlink Resource Allocation 
> 
> The Frame managemnt and the Downlink Resource Allocation  will be 
> implementd at the eNB.
> The eNB will be able to distribute the bandwidth between active downlink 
> flows Frame by Frame, TTI by TTI. Well-Know  Resource Allocation 
> Algorithms (such as Maximum Throughput and Proportional Fair) will be 
> implemented.
> 
> NOTE: the uplink scheduler will not be implemented during the GSoC. In 
> this primary step we should consider that uplink transmission (for 
> example CQI feedback and TCP ack) will be done without any allocation 
> algorithm, using a constant MCS. 
> 
> 
> Please, could you help me to improve my work plane?
> Thanks in advance!
> 
> Regards,
> 
> Giuseppe Piro  
>  
> 
> 
> [1] 3GPP TS 25.814 ( http://www.3gpp.org/ftp/specs/html-INFO/25814.htm ) 
> [2] Giuseppe Piro, Luigi Alfredo Grieco, Gennaro Boggia, and Pietro 
> Camarda", A Two-level Scheduling Algorithm for QoS Support in the 
> Downlink of LTE Cellular Networks", Proc. of European Wireless, EW2010, 
> Lucca, Italy, Apr., 2010 
> (draft version is available on 
> http://telematics.poliba.it/index.php?option=com_jombib&task=showbib&id=330 
> <http://telematics.poliba.it/index.php?option=com_jombib&task=showbib&id=330> 
> )
> [3] 3GPP R1-081483 (available on 
> http://www.3gpp.org/ftp/tsg_ran/WG1_RL1/TSGR1_52b/Docs/R1-081483.zip)
> 
> 
> 
> -- 
> Giuseppe Piro
> Ph.D. Student
> DEE - Politecnico di Bari
> v. Orabona 4, 70125 - Bari, Italy
> e-mail:    g.piro at poliba.it <mailto:g.piro at poliba.it>
>               peppe.piro at gmail.com <mailto:peppe.piro at gmail.com>
> Phone:   +39 0805963301
> Web:       http://telematics.poliba.it/piro