| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Hi,dear attendees
Just like a chinese idiom said that Mending
the hurdle after losing sheep is not late yet.
To avoid too much disagreement, I request
a discussion. Please give me the response to the contribution
"clarification on link level simulation" before the meeting if you have any time
that I can take in your travel. So I can understand your
opinion clearly and prepare some thing suitable for the meeting.
Fortunately, the contribution is not too long. Here
it is.
Introduction Link level simulation has the ability to compare different design without the unnecessary interference. So link level simulation can give determined conclusion in some AI technique itself. Further, the results from the link level simulation are the necessary input of system simulation. In this document, we present the necessary link level simulation modeling assumptions and clarify what we can achieve from link level simulation. Link level Simulation
Modeling 1. Algorithm selection The link simulation modeling assumptions shall include the necessary part of transmitting and receiving structures in a system in order to compare the performance of different candidate algorithms. The evaluation may be measured by BLER or BER vs. EbN0. For example, channel coding and decoding algorithm, modulation and demodulation algorithm, etc can be selected based on link simulation. On contrary to system simulation modeling assumptions, the link simulation modeling assumptions shall be simple enough to avoid the unnecessary interference by other undetermined factors. For example, delay and channel estimations can be assumed to be ideal when we compare the performance of different channel coding algorithm. As a result of link simulation for algorithm selection, the margin of algorithm performance loss in system can be estimated. For example, the SNR loss due to realistic channel estimation can not be neglected in uplink link simulation because of poor pilots in uplink. 2. MC modes selection Link level
simulation can also be used to guide AMC design. We can simulate various
Modulation & Coding sets, and select suitable subsets as reference MC modes
to support adaptive MC. 3.
Common channel
design Common control channels (such as Broadcast channel, SYNC
channel, etc) should serve most of users in the cell, including those at the
cell boundary. On the other hand, performance of the common control channel will
restrict the maximum cell coverage. Typical channel environment (cumulative distribution
function of the SNR, for example) can be obtained through system level
simulation. This ˇ°SNR distributionˇ± has considered about path loss, shadow
fading, inter-cell and intra-cell interference, etc. We can design our common
channels to perform well in the typical channel environment. That is, common channels design can be evaluated
based on the SNR distribution obtained from system level simulation.
4.
Link-system simulation
interface The link
simulation modeling assumption shall satisfy the requirement of system level
simulation. If
HARQ is used, BS scheduling will consider the block error of the received packet
(downlink, for example). In fading channels, coding gains of different
transmission times are different. The performance of each retransmission depends
on the coding/decoding scheme. The BLER vs. SNR curves for the first
transmission and the BLER vs. SNR curves for the second transmission are
different. All the curves should be known before system level simulation. Since
it is impractical to simulate all the combinations of various channel models,
transmission times and MC set, we can select some typical subsets.
We
should also define how to get ˇ°BLER vs. instantaneous SNRˇ± curves. These curves
are more suitable to a system employing link adaptation than ˇ°BLER vs. average
SNRˇ± curves. Further, the link level simulation results
for channel models with typical AS and 5.
Control channel Signaling is very important to guarantee traffic
transmitting. If
an ACK is interpreted as NACK, the correct packet will be retransmitted and
throughput decreases. If a NACK is interpreted as ACK, the error packet will not
be retransmitted and the packet is lost. So NACK is more important than ACK. We
can give NACK more power or more duplicated transmission than ACK to ensure its
performance. Suitable power offset or transmission times can be simulated by
link level simulation or only by theoretical analysis.
Other control channels, such as scheduling, channel
quality feedback, may use robust modulation & coding scheme or use different
transmitting power compared to traffic channel to guarantee their
performance. 6.
Pilot
channel Optimal pilot channel power setting should be simulated.
The total power is fixed, if the proportion of pilot channel is high, channel
estimation will be more correct, but at the same time, traffic power will
decrease and impact traffic SNR. On the other hand, if the proportion of pilot
channel is low, traffic power will increase, but at the same time, channel
estimation will be distorted, and impact traffic demodulation. So the optimal
ratio between pilot channel and traffic channel should be considered. Conclusion The link level simulation modeling assumptions which are independent of candidate technique shall be included in the evaluation criteria.
Best regards
Bingyu Qu
|