A Software Architecture for Simulating IEEE e HCCA

Transcription

1 A Software Architecture for Simulating IEEE e HCCA Claudio Cicconetti, Luciano Lenzini, Enzo Mingozzi, Giovanni Stea Dipartimento di Ingegneria dell'informazione University of Pisa, Italy IPS-MoMe 2005 Warsaw, Poland, th March 2005

2 Summary Introduction Software Architecture Implementation and preliminary results Conclusions 2

3 Introduction New (EuQoS) applications require QoS videoconference VoIP online gaming... The legacy IEEE lacks QoS support Distributed Coordination Function Point Coordinated Function Solution: e Enhanced Distributed Channel Access (prioritized access) HCF Controlled Channel Access (parametrized access) 3

4 Legacy DCF Distributed protocol, based on CSMA/CA: listen before transmit collisions detected with positive acknowledgment binary exponential backoff procedure if collision All the stations access the medium with the same procedure Only suitable for best-effort traffic 4

5 Legacy PCF Contention-Free Periods (CFPs) alternates to Contention-Periods (CPs) at fixed intervals During CFPs the AP cyclically polls the STAs Still unsuitable for providing QoS 5

6 802.11e Enhancements QoS Access Point (QAP), with enhanced scheduling capabilities Admission control QoS Stations (QSTAs), capable of Contention-based access DCF, EDCA Responding to polls PCF, HCCA 6

7 802.11e - EDCA Distributed protocol Up to four Access Categories Based on the differentiation of the CSMA/CA parameters Achieves relative differentiation of traffic from different ACs If in infrastructure mode, admission control at the QAP transmit queues per-ac AC 1 Link Layer Mapping (MSDU, UP) to Access Categories AC 2 AC 3 AC 4 internal collision resolution wireless medium access 7

8 802.11e HCCA (i) Centralized Up to 8 Traffic Streams (TSs) for each QSTA Traffic specs and requirements (TSPEC) negotiation Mean data rate, delay bound, etc. Admission control of TSs at the QAP Absolute QoS guarantees enforced by scheduling Controlled Access Phases (CAPs) 8

9 HCCA Scheduling TXOP i Service Interval (SI) TXOP j HCCA TXOP k DCF EDCA TXOP i TXOP j TXOP k The e does not define a standard scheduling algorithm at the QAP A reference scheduler is provided TDM-like scheduling: TXOP of fixed duration at fixed time intervals for all admitted TSs Different schedulers are currently being proposed in the literature We need a common framework for evaluation and testing 9

10 Software Architecture Modular architecture MAC fully compliant to the standard HCCA function MAC functions separated from HCCA scheduling Allows for flexible integration of various schedulers outgoing packets Upper Layers incoming packets Measurement Classifier outgoing packets Medium Access Control HCCA Scheduler 10

11 MAC (i) MAC is implemented as a FSM driven by a set of events CAP_Hand Receive/Data_Receive Has_Control/Lost_Control Success Transmit 11

12 MAC (ii) 12

13 HCCA Scheduler (i) Listens to a subset of the MAC events Has_Control/Lost_Control Possibly others, depending on its actual needs General interface: enque(): adds a new packet coming from the LL deque(): pops the HOL packet QAP-specific functions get_next_cap(): returns the expiration time of the next CAP addtspec(): requests the admission of a new TS 13

14 14 HCCA Scheduler (ii) QSTA DEQUE do/ dequeue() do/ get_queue_size() BUSY IDLE enque() QAP ACTIVE ACTIVE DEQUE do/ dequeue() HCCA_LOST_CONTROL addtspec() enque() HCCA_HAS_CONTROL HCCA_LOST_CONTROL BUSY IDLE HCCA_STOP HCCA_START NEXT CAP do/ get_next_cap() HCCA_STOP HCCA_START HCCA_HAS_CONTROL

15 Implementation The software architecture was implemented in the Network Simulator 2 simulation environment QAP scheduler: reference e QSTA scheduler: FIFO with only one TS QSTA 1 TS n TS 1 TS 2 TS n... TS 1 TS 2 TS n QSTA 2 TS n QAP QSTA n TS n 15

16 Preliminary Results (i) e QSTAs legacy STAs e QAP 1 QAP 3 QSTAs (bidirectional video streaming session) 3 legacy STAs (asymptotic condition) 11 Mbps Error-free channel No RTS/CTS, no MAC fragmentation 16

17 Preliminary Results (ii) Maximum delay (s) EDCA video only Maximum delay of video flows EDCA data+video uplink downlink HCCA data+video Time (s) 17

18 Preliminary Results (iii) Throughput (bps) data (throughput) data (collisions) video downlink video uplink Collision probability (%) Time (s) 18

19 Conclusions and future work A software architecture for simulating the IEEE e HCCA was defined The scheduling and MAC functions were decoupled by a generic communication interface The contributed framework has been implemented using ns2 Future (ongoing) work Defining, implementing and comparing different HCCA scheduling algorithms for e 19

20 End of presentation Thanks for your attention?? // 20