MACAW: A Media Access Protocol for Wireless LAN's

Summary:
The authors in this paper propose MACAW that modifies MACA. Techniques that have been proposed for wireless communication include:

• CSMA (Carrier Sense Multiple Access) that senses carrier before sending data. this faces "Hidden node" and "exposed node" problem.
• MACA (Multiple Access Collision Avoidance) uses RTS-CTS-Data along with exponential backoff to select retransmission time.

MACAW proposes four changes:

1. Fairness:
   
• The binary exponential backoff (BEB) algorithm does not provide fairness. It can happen that a terminal with low backoff counter will always win over another terminal with higher backoff counter where both terminals are sending data at channel capacity. MACAW proposes to have all terminals in a range update their backoff counter on hearing a packet.
  
• To make BEB algorithm not have large oscillations the authors propose MILD technique.
• Fairness over stream instead of per station is achieved by implementing separate queues for each stream.
2. RTS-CTS-DS-DATA-ACK modifications:
• RTS-CTS-DATA-ACK exchange provides error checking at link level.
• Send DS before DATA so that all nearby terminals wait until ACK is heard. This provides efficient synchronization of information about contention periods.
  
• RRTS solves contention problem due to lack of synchronization. Example we have A-to-B and D-to-C stream where B and C are close to each other. In this case D-to-C stream RTS may not be responded because A-to-B stream has large data size and short gaps between complete data transmission as well as completion of B's next CTS. In order to solve this problem RRTS can be issued by C to D (as response to non-responded RTS) during the next contention period in order to initiate communication .
  
3. Location dependent
• Since congestion varies per region it is required to modify BEB to be per sender-receiver pair. This is achieved by having the backoff scheme copy both sender and receiver backoff counters.

Background:
It is helpful to read about 802.11 MAC protocol before reading this paper.

Impact and Critique:
The paper very nicely gives a brief overview over the past protocols and their drawbacks. I felt that per stream implementation by keeping separate queue for each stream was similar to the concept of VOQ.

MACAW protocol addresses a lot of issues like fairness and better congestion control but there are issues like NACK and multicast that have not been addressed or do not have a good solution. While reading the paper I felt that few modifications were made to handle very specific type of traffic and may have resulted in making the protocol very complicated. Moreover the experiments are run on a simulator thus the scenario may not be exactly the same as real-time traffic. The MILD scheme increases the counter interval by 1.5 times and decreases by 1. There is no mention about how these parameters were chosen. Do they need to be changed as per traffic pattern?