Review 15 (better formatting)

From: Charles Reis (creis@cs.washington.edu)
Date: Mon Nov 22 2004 - 01:26:56 PST

  • Next message: Danny Wyatt: "MACAW: A Media Access Protocol for Wireless LAN's"

    MACAW: A Media Access Protocol for Wireless LANs
    Bharghavan, Demers, Shenker, Zhang, 1994.

    The authors simulate and evaluate an early proposal for media access protocols for wireless networks, and they design several enhancements to improve overall throughput and fairness of allocation in noisy and congested environments. Discounting token-based approaches for the time being, they start with MACA (Multiple Access, Collision Avoidance), which uses RTS/CTS packets to avoid the hidden terminal problem and binary exponential backoff to recover from congestion.

    The paper uses a series of worrisome corner cases to illustrate problems that MACA does not solve, and it proposes relatively straightforward fixes that are shown through simulation to improve behavior in the relevant case. These fixes include a gentler adjustment in the backoff algorithm, per-packet ACKs, more synchronization info with DS and RRTS (request-to-request-to-send) packets.

    However, the quick fixes do not appear to take a global view of the problem, occasionally causing problems among each other (such as the new backoff algorithm leading to high backoff counters) or suggesting additional measures ad infinitum (RRRRRTS?). There are many scenarios shown which are still unsolved (and likely many more that are not shown), suggesting this approach leaves much to be desired.

    More importantly, many of the assumptions seem naive and extremely limited to the "nanocell" model with very low interference and symmetric links. The observation that two nearby senders can transmit at the same time if their receivers only hear one of them might be problematic if the senders RF signals interfere with each other near the senders. Similarly, highly noisy environments and larger reception areas likely suggest different priorities and solutions for MAC protocols.

    Overall, some of the measures suggested still represent reasonable improvements over earlier MAC protocols, and RTS/CTS with per-packet ACKs are still in use in some 802.11 implementations today. A better understanding of true RF propagation and interference, along with a more global view of the MAC problems, might help to improve protocols for the future.


  • Next message: Danny Wyatt: "MACAW: A Media Access Protocol for Wireless LAN's"

    This archive was generated by hypermail 2.1.6 : Mon Nov 22 2004 - 01:27:02 PST