1. Block error(BLER) target
Evaluation of Dynamic
Block error(BLER) target
Selection in HSUPA
20th November, 2014
Tejas Subramanya
Conducted in: Nokia Networks
Supervisor: Prof. Riku Jantti
Instructor: Rua Philippe

2. Outline Introduction Objectives of the thesis Power control procedures
‘Dynamic BLER’ feature
Simulation procedures
Simulation results
Problems and solutions when the feature is active
Conclusion

3. Introduction (1) HSPA -> High Speed Downlink/Uplink Packet Access
Why WCDMA/HSPA ?
HSPA -> High Speed Downlink/Uplink Packet Access
Shared resources -> DL : Channelization codes; UL : Interference
HSUPA – Dynamic HSUPA BLER target selection in outer loop power control to reduce interference

4. Introduction (2) Uplink channels Physical channels:
E-DPDCH -> Carries data over air interface
E-DPCCH -> Carries control information related to E-DPDCH
DPCCH -> Carries transmit power control commands
HS-DPCCH -> Carries acknowledgements and CQI information
Traffic channels:
E-DCH -> Carries HSUPA related data
DCH -> Carries WCDMA related data

5. Objectives of the thesis
Understand the importance of ‘Dynamic HSUPA BLER’ feature
Simulate the feature based on system test logs
Study the problems under low traffic and high traffic within the cell
Discuss the solutions to each of the problems with simple simulations
Analyze the gain achieved in terms of cell throughput based on simulation results

6. Power Control Procedures (1)
Power control – Immediate response for fast changes in signal and interference levels.
Inner loop power control (ILPC)
- Control uplink transmission powers of UE to reach target SIR
Outer loop power control (OLPC)
- Calculate minimum target SIR required for sufficient quality of connection (Ideal BLER target fixed to 10% )

7. Power Control Procedures (2)
Outer Loop Power Control
- OLPC entities (1 per traffic channel)
- OLPC controller (1 per call connection)
- OLFTOR
OLPC entity:
- Measures BLER
- Reports it to OLPC controller
OLPC controller:
- Controls the state of OLPC entities
(active, semi-active or inactive)**
- Adjusts target SIR
OLFTOR:
- Forwards this information to Layer 3
(Handover control in case of quality deterioration)

8. Dynamic BLER feature (1)
Classification of users into different traffic types in OLPC
Inputs: Frame rate, FP bit rate, Number of HARQ retransmissions
Peak traffic type (Low BLER target)
- Condition: Percentage of UE packets in FP frame > 90%
Ideal BLER target: 8%
Bursty traffic type (Moderate BLER target)
Condition: FP frames per second <= 10 (10ms TTI); FP frames per second <= 50 (2ms TTI)
Ideal BLER target: 10% for 0 HARQ retransmissions
Continuous traffic type (High BLER target)
Condition: FP frames per second > 10 (10ms TTI); FP frames per second > 50 (2ms TTI)
Ideal BLER target: 10% for 1 HARQ/31.6% for 0 HARQ retransmissions

9. Dynamic BLER feature (2)
Algorithm 1: Mixture of DCH and E-DCH traffic channels in UE
Determine dynamic BLER target using below equations
DCH:
E-DCH:
Determine the change in SIR required
Calculate the target SIR
- If all transport blocks of the frame are ok, ∆SIR = sir_down_step_size
- If any transport block is CRC erroneous, ∆SIR = sir_up_step_size 1 2

10. Dynamic BLER feature (3)
Algorithm 2: Only E-DCH traffic channels in UE
Determine the change in SIR required.
BLER estimation = 0, if frame is ok;
BLER estimation = 1, if frame failed;

11. Determine traffic type and ideal BLER target
Simulation Procedure (1)
‘Noise rise’ gives the total uplink load factor (capacity of the cell).
Load factor per user= Sum(DPCCH,HS-DPCCH,E-DPCCH,E-DPDCH) load factors * activity factor of the user
HARQ Rx number (Input)
FP bit rate (Input)
Frame rate
(Input)
Determine traffic type and ideal BLER target
Measured BLER
(Input)
Calculate SIR target
DPCCH load factor
HS-DPCCH load factor
E-DPCCH load factor
E-DPDCH load factor

12. Simulation Procedure (2)
Load factor DPCCH:
LDPCCH = 10^ ((SIR – 10log10 (SF))/10) , where
‘SF ‘ is the spreading factor of DPCCH.
‘SIR’ is the signal to interference ratio of the user.
Load factor HS-DPCCH:
LHS DPCCH = 10^ ((10log10 (PO) + 10log10 (LDPCCH))/10) , where
‘(βhs/βc)’ is given by quantized amplitude ratio table for HS-DPCCH based on the signalled values. 3 4

13. Simulation Procedure (3)
Load factor E-DPCCH:
LE-DPCCH = LDPCCH * Aec2 , where
‘Aec’ is the gain factor for E-DPCCH (βec)/ gain factor for DPCCH (βc ).
‘Aec’ is obtained by quantized amplitude ratio table for E-DPCCH based on the signalled values.
Load factor E-DPDCH:
LE-DPDCH = LDPCCH * Absolute grant value
Absolute grant value = Aed*Number of codes
‘Aed’ is the ’Gain factor for E-DPDCH (βed)/ gain factor for DPCCH (βc ).
‘Absolute grant values’ are obtained by scheduling grant tables based on E-DCH transport block size used in uplink transmission. 5 6

14. Simulation Procedure (4)
Average user throughput and cell throughput are calculated based on the activity factor, bearer bit rate and number of users in the cell.
Average user throughput (number of users) = (Sum (Activity factors of all the users)*Bit rate of radio bearer)/number of users
Cell throughput (number of users) = Average throughput (number of users) * number of users 7 8

15. Feature inactive:841 Kbps
Simulation Results (1)
10 HSUPA users within the cell (10ms TTI)
Feature inactive:841 Kbps
Feature active:
1.24
Kbps

16. Simulation Results (2) 10 HSUPA users within the cell (2ms TTI)
Feature inactive:
1.7 Mbps
Feature active:
2.4 Mbps

17. Problems and Solutions when the Feature is active (1)
HSUPA users within the cell < 3
Problem: End to end TCP downlink throughput reduces if the traffic type of the user is continuous.
Reason: If user is continuous, high BLER target is used which results in low SIR target, which in turn results in the increase of uplink HARQ Retransmissions.
Increase in HARQ retransmissions leads to increase in RLC retransmissions and thus increases TCP round trip time. Thus, TCP ACK’s in uplink are delayed and TCP downlink throughput reduces significantly.

18. Problems and Solutions when the Feature is active (2)
HSUPA users within the cell < 3
Simulation proof:
The graph shows the behaviour
Of TCP round trip time and
TCP downlink throughput
w.r.t BLER.
Solution: Feature is deactivated when the number of users within the cell is less than 3.
TCP RTT
TCP throughput

19. Problems and Solutions when the Feature is active (3)
2. HSUPA users within the cell > 35
Problem: All the users tend to behave more like bursty traffic type even though the users are transmitting continuously. This results in the use of moderate BLER targets of bursty traffic type even for continuous users and thus reduces the gain achieved in cell throughput.
Solution: Users are not differentiated into continuous and bursty traffic type above 35 users and only BLER targets of continuous users are used. This improves the gain achieved in cell throughput.

20. Problems and Solutions when the Feature is active (4)
2. HSUPA users within the cell > 35
Before the problem is solved:
200 Kbps
After the problem is solved:
450 Kbps

21. Conclusion
‘Dynamic HSUPA BLER’ is a feature which reduces the uplink interference and thus increases the cell capacity.
Two shortcomings of the feature are addressed and the solutions are provided to overcome them.
- In low HSUPA traffic case, TCP downlink throughput decreases and thus feature is not activated.
- In high HSUPA traffic case, continuous users are treated as bursty and moderate BLER targets are used and thus user type differentiation is not done above 35 users. All the users are treated as continuous.