= Some discussion and material on Control Channel based 802.11 design =
----
P. Kyasunur, J. Padhye and P Bahl, "On the efficacy of separating control and data into different frequency bands",
----
== Key Points ==
 * Using separate frequency (lower range) and low data rate link for control channel message exchanges 
 * Channel access contention and reservations done on the control channel
 * Adjustible ''k'' factor, prereservation for upto ''k'' packets, depending on how much time it takes to transmit data packet.
 * Preservation upto ''k'' packets smoothens out the packet transmission time differences
 * The benefits of using control channel versus packet size of the data packet are shown in following Figure [[Image(ccm.PNG)]] 

== Effect of packet size ==
Thus, for every control channel bandwidth, there is some optimum packet size threshold, such as the benefits of using control channel become significant. Since application level generates packet sizes that suit the application, in order to create this "optimum packet size" many such packets are aggregated together.

== Effect of control channel range ==
 * By using a control channel range that is close to the interference range of the data channel, all nodes in the interference region can be notified of the impending transmission, thereby preventing data packet collisions.
 * If the transmission range of the control channel is too large, then contention resolution process on the control channel will reserve an unnecessarily large area, reducing spatial reuse.

== Protocol Architecture ==
The protocol architecture is shown in the following Figure 
[[Image(cca.PNG)]]

== Some Comments ==

 * Previous schemes only suggest two channels, one radio. This C2M needs 2 radios.
 * The performance improvement comes from a "shaky" assumption: RTS/CTS is sending in lowest rate. If RTS/CTS is in the same rate as DATA frames. The conclusion would be: little throughput improvement. 
 * RTS/CTS is not mandatory in IEEE 802.11. In most scenarios, the benefits of RTS/CTS are fairly small. If the CS range is set to equal to interference range, the hidden node could be avoided without RTS/CTS. 
 * Continue from above, is per-packet reservation a bad idea?
 * With a new digital channel, more comprehensive information could be transported in that channel, instead of per-packet reservation.
 * It is possible to put ACK in control channel for better spatial reuse (for receiving block problem....)