Topic: Wireless MAC (09/21/2006)
Paper: MACAW: A Media Access Protocol for Wireless LAN's (ACM SigComm 1994)
Reviewer: Eunkyu Lee


This paper develops a new media access protocol for wireless LAN based on MACA (Multiple Access Collision Avoidance), and addresses basic performance and design issues of the proposed the MACAW. As three background reviews, the authors describe Xerox Corporation's 'Nano-Cellular Radio Network' providing a nanocell radio environment consisting of pads and base stations, Carrier Sense Multiple Access (CSMA) providing the simplest carrier sensing model but having problems of 'hidden terminal' and 'exposed terminal', and Multiple Access Collision Avoidance(MACA) introducing RTS-CTS control packets to solve the problems in CSMA. When designing MACAW, the paper pays attention to the backoff algorithm and the RTS-CTS message exchange in order to achieve the high network throughput and the fairness in media access in wireless LAN. In terms of the backoff algorithm, the paper proposes a multiplicative increase and linear decrease (MILD) avoiding wild oscillation of backoff counter by decreasing the size of backoff interval, which provides an adequate level of fairness on resource usage. With respect to the message exchange, the paper proposes three new control messages added into the basic RTS-CTS-DATA message exchange used in MACA. At first, an acknowledge packet, ACK, is sent from the receiver to the sender when the receiver completes the data reception, which increases the reliability at the link layer under and advances the throughput performance. The next two control message are a Data-Sending packet (DS) sent by the sender just before sending a DATA packet and a Request-for-RTS (RRTS) sent by the receiver during the next contention period when receiving an RTS but being unable to respond immediately due to deferral. These two control packets provide synchronization information regarding the contention period, which enhances the fairness of access to the media. When evaluating the MACAW, two values of the network throughput and the resource balance are measured. Additional measurements regarding the total channel capacity are done in order to quantify the overhead burned by the MACAW protocol. The paper shows that the proposed MACAW outperforms the MACA in terms of improving fairness almost without sacrificing the network throughput. 

 

One of the major contributions of the paper is its investigation and analysis of problems in MACA. Based on the endeavour, it gives four key insights as follows. (1) The relevant contention is at the receiver, not the sender; (2) the congestion is location dependent; (3) The media access protocol should provide congestion information explicitly; (4) the media access protocol should provide synchronization information about the contention periods. Through the observations, the paper raises such problems why the MACA cannot be applied into the wireless environment and what modifications should be addressed. Throughout the paper, the authors focused on the backoff counter and the unfairness of resource access as improvement issues, and developed the proposed solution against these problems with simple scenarios covering various communication use cases. 

 

However, the paper has some weak points in terms of unclear criteria used in its solution and impractical scenario. The MILD scheme increases the counter interval by 1.5 times upon a collision and decreases the counter by 1 upon success. And the authors show that the scheme works well with the scenario presented in figure 3. However, what will happen if it increases by 1.6 times and decreases by 1.1? What proves the selected values are the optimal? The authors should give more refinement and explanation or present experimental results comparing these values. Even though simple scenarios help readers understand the problems and the solutions, they make the experiments to be seemed a laboratory simulation, not practical evaluation. Even the last two configurations simulate a part of the laboratory at PARC, which can degrade the fidelity of the results. 

 

In summary, the paper gives a significant perspective regarding the media access protocol in wireless LAN environment even though it has small shortcomings. It identifies problems inherent in MACA when applying for wireless LAN and proposes appropriate and simple solutions with minimum performance degradation. However, the paper can be improved by including clearance in ideas and doing experiment in the field.