Sunteți pe pagina 1din 20

Helsinki University of Technology

Communication Laboratory

Diploma Thesis

LINK ADAPTATION IN GENERAL PACKET RADIO SERVICES (GPRS) Juan Li


Instructor: Boris Makarevitch Supervisor: Sven-Gustav Hggman
Communication Laboratory, HUT
28th.May, 2002 Link Adaptation in GPRS 1/20

Helsinki University of Technology


Communication Laboratory

Important Role of GPRS

GPRS is being rolled out by operators around the world as the first vital step toward 3G.

Communicated GPRS launches 2000- December 2001

28th.May, 2002

Link Adaptation in GPRS

2/20

Helsinki University of Technology


Communication Laboratory

Data Flow in GPRS

Data flow and Segmentation Between the Protocol Layers in the MS

28th.May, 2002

Link Adaptation in GPRS

3/20

Helsinki University of Technology


Communication Laboratory

Four Coding Schemes of GPRS

Radio Block Structure for CS1 to CS3

Radio Block Structure for CS4

28th.May, 2002

Link Adaptation in GPRS

4/20

Helsinki University of Technology


Communication Laboratory

Link Adaptation in GPRS

Why link adaptation?


During the evolution to 3G, Link Adaptation (LA) has been identified as a key technology for evolved GSM systems, such as, AMR, GPRS, EDGE.

What is link adaptation?


Select the most suitable coding scheme accroding to the current channel quality.

Objective:
The objective of link adaptation is to achieve maximum throughput by selecting a suitable coding scheme instantaneously
28th.May, 2002 Link Adaptation in GPRS 5/20

Helsinki University of Technology


Communication Laboratory

Performance of Four Coding Schemes


Throughput vs. CIR
25

25

20 Throughput (Kbps) Throughput (Kbps)

20

15

15

10 CS1 5 CS2 CS3 0 0 5 10 15


CIR (dB)

10 CS1 5 CS2 CS3 0 CS4 0 5 10 15


CIR (dB)

CS4 20 25 30

20

25

30

without frequency hopping, user speed = 1m/s


25

without frequency hopping, user speed = 15m/s


25

20 Throughput (Kbps) Throughput (Kbps)

20

15

15

10 CS1 5 CS2 CS3 0 0 5 10 15


CIR (dB)

10 CS1 5 CS2 CS3 0 CS4 0 5 10 15


CIR (dB)

CS4 20 25 30

20

25

30

with frequency hopping, user speed = 1m/s 28th.May, 2002

with frequency hopping, user speed = 15m/s 6/20

Link Adaptation in GPRS

Helsinki University of Technology


Communication Laboratory

Link Adaptation Algorithms (1)

Two algorithms have been discussed in this thesis: --BLER-based algorithm: the parameter used to estimate the channel quality is the Block Error Rate (BLER). The BLER at intersections are taken as the thresholds and they are calculated as: Thr BLER = 1 Thr
max

--CIR-based algorithm: the parameter used to estimate the channel quality is the Carrier-to-Interference (CIR) instead. The thresholds can be directly taken from above figures based on the intersections.
28th.May, 2002 Link Adaptation in GPRS 7/20

Helsinki University of Technology


Communication Laboratory

Link Adaptation Algorithms (2)

An Example of Coding Scheme Updating for BLER-Based algorithm

28th.May, 2002

Link Adaptation in GPRS

8/20

Helsinki University of Technology


Communication Laboratory

Throughput Intersections vs. User Speed

Throughput intersection (dB)

28 24 20 16 12 8 4 1m/s

CS1&CS2 CS2&CS3 CS3&CS4

3m/s

5m/s

7m/s

9m/s

11m/s

13m/s

15m/s

MS speed, noFH

without frequency hopping 28th.May, 2002 Link Adaptation in GPRS 9/20

Helsinki University of Technology


Communication Laboratory

Simulation Model

Link level model for link adaptation in GPRS

28th.May, 2002

Link Adaptation in GPRS

10/20

Helsinki University of Technology


Communication Laboratory

Comparsion Under Ideal Situation


Throughput vs. CIR
23.0

23.0 22.5

BLER Throughput (Kbps) CIR

22.0 21.0 20.0 19.0 18.0 17.0 16.0 15.0

Throughput (Kbps)

22.0 21.5 21.0 20.5 20.0 19.5 12 14 16 18 20 22 24 26

BLER CIR

12

14

16

18

20

22

24

26

CIR (dB)

CIR (dB)

without frequency hopping, MS speed = 1m/s


23.0

without frequency hopping, MS speed = 15m/s


23.0

BLER
22.0

22.0

BLER CIR

Throughput (Kbps)

21.0 20.0 19.0 18.0 17.0 12 14 16 18 20 22 24 26

Throughput (Kbps)

CIR

21.0 20.0 19.0 18.0 17.0 16.0 15.0 12

14

16

18

20

22

24

26

CIR (dB)

CIR (dB)

with frequency hopping, MS speed = 1m/s 28th.May, 2002

with frequency hopping, MS speed = 15m/s 11/20

Link Adaptation in GPRS

Helsinki University of Technology


Communication Laboratory

Comparsion Under Practical Situation (1)

without frequency hopping


13dB_1m/s 19dB_1m/s 25dB_1m/s 13dB_15m/s 19dB_15m/s 25dB_15m/s mean_Thr. Std. BLER1 19.78 22.26 22.61 10.95 13.30 17.12 BLER1 17.67 4.79 BLER15 16.79 16.88 16.88 15.70 16.84 16.88 BLER15 16.66 0.47 BLERm 19.78 22.26 22.61 11.03 13.39 17.07 BLERm 17.69 4.75 CIR1 20.14 22.25 22.61 1.54 9.93 16.93 CIR1 15.57 8.31 CIR15 16.80 16.88 22.61 15.74 16.82 17.00 CIR15 17.64 2.48 CIRm 16.80 22.22 22.63 15.74 9.98 17.04 CIRm 17.40 4.67 fixedCS1 11.39 11.39 11.39 11.38 11.39 11.39 fixededCS1 11.39 0.00 fixedCS4 20.21 22.26 22.59 1.48 10.10 17.23 fixededCS4 15.65 8.33

with frequency hopping


13dB_1m/s 19dB_1m/s 25dB_1m/s 13dB_15m/s 19dB_15m/s 25dB_15m/s mean_Thr. Std. BLER1 17.61 20.98 22.09 10.04 17.96 22.12 BLER1 18.47 4.58 BLER15 16.60 16.88 16.88 15.98 16.86 16.88 BLER15 16.68 0.36 BLERm 17.61 20.98 22.09 10.04 17.96 22.12 BLERm 18.47 4.58 CIR1 18.38 21.28 22.09 0.14 18.44 22.17 CIR1 17.08 8.48 CIR15 16.58 21.21 22.08 15.97 18.46 22.22 CIR15 19.42 2.79 CIRm 16.58 21.21 22.08 15.97 18.46 22.22 CIRm 19.42 2.79 fixedCS1 11.39 11.39 11.39 11.35 11.39 11.39 fixedCS1 11.38 0.02 fixedCS4 18.57 21.26 22.08 0.12 18.50 22.14 fixedCS4 17.11 8.48

28th.May, 2002

Link Adaptation in GPRS

12/20

Helsinki University of Technology


Communication Laboratory

Comparsion Under Practical Situation (2)

without frequency hopping


20 18 16 14 12 10 8 6 4 2 0 BLER1 BLER15 BLERm CIR1 CIR15 CIRm Link Adaptation Schemes mean Thr. Std.

28th.May, 2002

Throughput (Kbps)

Link Adaptation in GPRS

13/20

Helsinki University of Technology


Communication Laboratory

Comparsion Under Practical Situation (3)

with frequency hopping


20 18 16 14 12 10 8 6 4 2 0 BLER1 BLER15 BLERm CIR1 CIR15 CIRm Link Adaptation Schemes mean Thr. Std.

28th.May, 2002

Throughput (Kbps)

Link Adaptation in GPRS

14/20

Helsinki University of Technology


Communication Laboratory

Comparsion of Link Adaptation and Fixed Coding Scheme

without frequency hopping

with frequency hopping

28th.May, 2002

Link Adaptation in GPRS

15/20

Helsinki University of Technology


Communication Laboratory

Remarks on Statistic-Based Method

The above discussions proved that the proposed statisticbased method offered a powerful tool in evaluating and optimizing link adptation scheme. It is interesting to note that the method is flexible. More accurate model can be obtained by increasing the number of typical circumstances if necessary. The method is also open to other link adaptation schemes of interest.

28th.May, 2002

Link Adaptation in GPRS

16/20

Helsinki University of Technology


Communication Laboratory

Preliminary Explanation of Superior Performance of CIR15 (1)

20 Throughput difference (Kbps) 1m/s CS3-CS4 15 10 5 0 0 -5 -10 CIR (dB) 5 10 15 20 25 30 15m/s CS3-CS4

Throughput difference between CS3 and CS4 without frequency hopping, MS speed = 1m/s.

28th.May, 2002

Link Adaptation in GPRS

17/20

Helsinki University of Technology


Communication Laboratory

Preliminary Explanation of Superior Performance of CIR15 (2)

Throughput loss resulted by misjudgement

28th.May, 2002

Link Adaptation in GPRS

18/20

Helsinki University of Technology


Communication Laboratory

Conclusion

qThe simulation suggested that CIR-based algorithm was superior to BLER-based algorithm under ideal situation. qA method based on statistical analysis was proposed to evaluate the performances of different link adaptation schemes for the first time. According to the comparison, CIR-based link adaptation scheme owns the best performance when thresholds are selected by assuming high user speed (15m/s). The method was proved helpful and powerful in evaluating and optimising link adaptation schemes. qA go-without-saying performance comparison for system either with link adaptation or with fixed coding scheme was also done. System with fixed coding scheme would have low throughput and possible high throughput variation. Therefore link adaptation was proved to be necessary. qA preliminary study was done to reveal why CIR15 was superior to other schemes by considering the throughput difference between different schemes. qThe discussion suggested an new topic for link adaptation study on how to the find the optimized threshold to improve the performance, which had been proved to be interesting and important.
28th.May, 2002 Link Adaptation in GPRS 19/20

Helsinki University of Technology


Communication Laboratory

Question?

28th.May, 2002

Link Adaptation in GPRS

20/20