IMA Wireless Workshop 6/05 – 1 Wireless Local Area Network Simulation Abstract: We describe a software system that simulates wireless local area networks supporting heterogeneous services and multiple protocols. We present applications of this system in three areas: analysis of voice capacity, maximization of data throughput while protecting voice quality of service, and the design and evaluation of scheduling algorithms in a polling-based system. Martin Eiger, Moncef Elaoud, David Famolari, Raquel Morera Telcordia Technologies Piscataway, NJ

IMA Wireless Workshop 6/05 – 2 Overview  Multi-year collaboration  Model and analyze many configurations – Existing standards – Possible future standards  Several IEEE variations – One queue/separate queues – Contention-based/contention-free  C/UNIX with CSIM simulation package

IMA Wireless Workshop 6/05 – 3 Wireless Local Area Network  Heterogeneous services  Shared transmission medium access point

IMA Wireless Workshop 6/05 – 4 Listen-Before-Talk Protocol  Transmit only after idle medium for appropriate IFS (interframe space) interval  Random backoff after every transmission – Pick random integer, decrement periodically, transmit when counter reaches 0 – Suspend countdown on busy medium, resume after idle IFS – Promotes fairness  Increase random backoff range after collision; restore on success

IMA Wireless Workshop 6/05 – 5 Simulator “Knobs”  Time to simulate  standard  Number of nodes  Mobility  Transmission rates  Service mix  Packet sizes  Packet arrival rates  Delay thresholds  Interframe spacing  Contention windows  Queue limits  Scheduling policy  Desired output ●

IMA Wireless Workshop 6/05 – 6 Voice Capacity (IEEE b) Conclusions:  QoS degrades for all users concurrently  Packetization interval is key determinant of max capacity (reason: fixed overhead per packet) packet- ization intervals:

IMA Wireless Workshop 6/05 – 7 Multiple Queues  Several queues at each node  Each queue senses medium, backs off independently  Motivation: set each queue’s interframe space and contention window parameters separately to assure quality of service

IMA Wireless Workshop 6/05 – 8 voice AIFSCW voiceUDPvoiceUDP Voice and Data  Conclusion: Failing to adapt AIFS and contention window values can cost ~20% of data throughput max throughput fixed AIFS and CW

IMA Wireless Workshop 6/05 – 9 Contention-Free Scheduling  Access point grants Transmission Opportunities Uplink Downlink poll pkt 1 … pkt n block ack req block ack pkt 1 … pkt n block ack req block ack  The big question: Scheduling

IMA Wireless Workshop 6/05 – 10 Scheduling Problem  Inputs – Traffic stream specifications (packet size, average packet arrival rate, …)  Output – AP schedules traffic streams – Polling-based packets meet delay bounds – Maximize time left over for contention- based traffic

IMA Wireless Workshop 6/05 – 11 A Work in Progress  We have examined several scheduling algorithms  Algorithms with best performance – Account for delays (be sure to grant TXOPs often enough) – Try to aggregate (don’t grant TXOPs too often)  Patent work begun; results to be published

IMA Wireless Workshop 6/05 – 12 Summary  Wireless LAN simulation – Many environments – “What if” scenarios  Results – Voice capacity – Maximize data while protecting voice – Scheduling protocol  Ultimate goal – Wireless LAN design and operations maximize capacity, assure quality of service