Summary

1. Summary • Motivation & Goals • Cdma2000 1xEV-DO: Overview • Traffic Model • EV-DO Model: Layers 1 e 2 • Experiments, Results and Discussion • Conclusions

2. Motivation & Goals • Evaluate EV-DO rev.0 as a alternative for the return channel of the Digital TV: • Throughput • Delay • User population • Fairness • Could we do better?

3. Pilot DRC ACK RRI Phase reference for demod.; Channel timing Forward channel data packets acknowledgment Reverse Rate Indicator Forward link rate sent to the BS Cdma2000 1xEV-DO: Overview • Reverse Link • Rates from 9.6 to 153.6Kbps per user • BPSK modulation • 53.30ms packets • Besides the data channel • Control Channel CDMA Power controlled: - open loop - closed loop

4. 1.67ms TDM slots Always at max. power Cdma2000 1xEV-DO: Overview • Forward Link • Rates from 38.4kbps to 2.4Mbps through adaptive coding/modulation • Depends on the state of the forward link • Higher bit rates: lower robustness to channel impairments • PFS recommended for slot allocation • Main goal: throughput

5. 1.67ms TDM slots Always at max. power Pilot RPC RA Bit open loop power control closed-loop power control congestion control Cdma2000 1xEV-DO: Overview • Forward Link (more...) • Four Packet Interlacing • Avoids temporary monopolization of slots • Early Termination • Gradual redundancy • Besides the data channel • Control Channel

6. Tangram-II Simulation Model Overview • Simulation Tool: • Tangram-II (Federal University of Rio de Janeiro – Brazil) • Why Tangram-II instead of NS? • Modeling is done through a high level interface: simplicity • Includes additional high-level constructs to facilitate the modeling task COPPE/UFRJ CETUC/PUC-RIO

7. Proposed model • Detailed (Physical and Link layers) EV-DO model • Related work typically: • Physical layer modeling • Lack of detailed traffic model • Few works modeling higher layers: • No open-source model presenting system capacity results as a function of user population • No detailed fairness analysis • Almost no clues on what could be done about fairness

8. Traffic Model • Assumptions • There is no users entering or leaving the system • No mobility – predominant scenario in the BDTvS

9. Traffic Model (more...) • Web users, on-off source COPPE/UFRJ CETUC/PUC-RIO

10. EV-DO Model • Assumptions: • A single EV-DO cell, no setorization: • Thus, no soft-handoff • Early Termination not implemented • Static users: low channel diversity

11. EV-DO Model: Physical layer • Propagation • Total power loss Ltotal[dB] propagation penetration penetration loss (10dB) Propagation loss shadowing fading (log-normal dist. de mean=0 and σ =8dB dense urban scenario)

12. EV-DO Model: Physical layer • Propagation (more...) • Okumura-Hata’s Propagation model for dense urban scenario BS antenna height (40m) dense urban correction factor : Carrier frequency (450MHz) AT height (1.5m)

13. EV-DO Model: Physical layer • Power Control: • Open Loop • Step 1: Pilot channel sensing • Step 2: Choose the lowest power such that: • After power losses it still reaches the receiver with enough strength to achieve the desired PER (1%)

14. EV-DO Model: Physical layer • Power Control (more…): • Closed loop • Step 1: BS calculates for each user • Step 2: • Matches the received power with d • Commands the AT to increase or decrease its power pilot channel received energy channel bandwidth (1.25MHz) thermal noise total perceived interfering power

15. EV-DO Model: Physical layer • DRC Estimation: Geometric Method • Step 1: Measure the received power from the BS: X • Step 2: Calculate SINR: Relation between X and the interference from rings 1 and 2 First interfering ring Second interfering ring

16. EV-DO Model: Physical layer • DRC Estimation : • Step 3: Find the higher DRC that matches the calculated SINR value, send it to the BS DRC Rate (kbps) SINR (dB) PER = 1%

17. EV-DO Model: Link layer • Scheduling Algorithm – PFS • Choose user j • Updating average rate for user i Last DRC value Received from user i User i average transmission rate User i current transmission rate PFS “fairness” parameter

18. EV-DO Model: Link layer • Congestion Control • Noise rise – δR • δR =Nt/N0 • δR = 5 as a threshold for the RA bit activation • If the base station activates the RA Bit, ATs decreases its reverse date rate transmissions thermal noise power total interfering power

19. Experiments: Considerations • Dense urban scenario • Web user population: from 10 to 80 • Interest Metrics: • Throughput; Delay; Fairness

20. Results: User Throughput vs. Pop. and zone • Decreasing Throughput with increasing population and distance Throughput (Kbps) Population Zone

21. Results: User Delay vs. Pop. and zone • Increasing delay with increasing population and distance Delay (s) Zone Population

22. alpha = 1.0 alpha = 0.001 Round Robin PFS throughput (Kbps) Zone Results: User Throughput vs. PFS α • PFS α - Extreme values: no significant fairness variation • α = 1 – Round Robin • α = 0.001 – PFS: Strong Throughput priority

23. Discussion: Fairness issue • Throughput, delay quite worse for the most distant users • PFS α parameter adjustment • Same fairness issue • Worst overall throughput

24. 1-zone 2-zones 1-user Proposed Solution: Directional Antenas • Former experiments: • Experiment 1: • Experiment 2: • Experiment 3: no-users (no directional antennas) … … … … … … … … … … 1 8 9 10 2 3 Distance Zone

25. Results: User Thoughput - fairness through directional antennas • Population of 60 users No user 1-user 1-zone 2-zones throughput (Kbps) Zone • From white to black bars: Increasing Fairness (thoughput)

26. Results: User Delay - fairness through directional antennas • Population of 60 users No user 1-user 1-zone 2-zones Delay (s) Zone • From white to black bars: Increasing Fairness (delay)

27. Results: User throughput - directional antennas for two zones 60 users 80 users throughput (Kbps) Zone directional antennas • Increased population: No significant fairness difference • The are no direction antennas in zones 1 to 8