Source : Dynamic Backoff for Wireless Personal Networks by Ai-Chun Pang and Hsueh-Wen Tseng
Authors propose a Memorized Backoff Scheme (MBS) with the Exponential Weighted Moving Average (EWMA) approach to dynamically adjust the size of contention windows.
[Fig – The Network Topology]
In the star topology, each mobile station communicates with others via the coordinator. The coordinator is usually located at the fix location, and directly connected to the socket for power supply. On the other hand, in the peer-to-peer topology, direct communications between mobile stations can be provided. The peer-to-peer topology is used for non-infrastructured wireless networks. The multihop routing is allowed in the peer-to-peer topology, and routing path can be dynamically updated
[Fig – The Superframe Structure]
A superframe structure for the beacon-enabled network. A superframe consists of 16 slots. In the 16-slot superframe, the first slot is used for the Beacon frame sent from the coordinator to mobile stations. The beacon includes the information for timing synchronization, system configuration, a list of the mobile stations that have to receive the data frames from coordinator and so on. The remaining 15 slots are divided into two parts. The first part is contention access period (CAP), and the second part is contention free period (CFP). In CAP, mobile stations access the medium by using contention. On the other hands, the slots in CFP are reserved for spesific mobile stations assigned by the coordinator. However, the coordinator is also in charge of the adjustment for the lengths of CAP and CFP.
A Memorized Backoff Scheme :
[Fig. – The Data Transmission Procedure]
If more than one mobile stations intend to send/receive the data frame, the CSMA/CA mechanism is used to solve the collision for the medium access. After listening to the beacon issued by the coordinator, the mobile stations is informed that a data frame at the coordinator is destined for the mobiloe station. After taht, the mobile station executes the several steps. The detail of these steps can be read in fullpaper :).
[Fig – The backoff Flows for IEEE 802.15.4 and MBS]
Detail explanation in fullpaper.
Simulation Model and Numerical Examples :
The length of dataframes is exponentially distributed with a mean of 1/mu UnitBackoffPeriods (slots)
Table System Parameters Used in the Simulation :
The transmission range of a coordinator is assumed to be 10 meters for the transmission rate of 250 Kbps.
[Fig – Simulation model]
- Goodput for all schemes (MBS,MBS+EWMA, IEEE 802.15.4) increases and then decreases as the traffic load increases, MBS and MBS+EWMA schemes achieve higher goodput than IEEE 802.15.4. The hig goodput for MBS and MBS+EWMA mainly results from the decreases of the number of the number contentions/collisions. In other words, by using MBS and MBS+EWMA, the backoff overhead is significantly reduced and therefore the goodput improves.
- The effects of the traffic load on the completions rate Rc [Rc = the number of the successful transmitted frames over the total number of transmitted frames]. Rc decreses as the traffic load is heavy.
- The effects of the traffic load on the average queuing delay and average MAC delay. The average queuing and MAC delay increase as the traffic load increases.
- The average number of Nc of collisions occured for each data frame prior to being successfully transmitted. Nc increases as the traffic load increases. The large Nc value indicates that with a data frame arrival, the mobile stations should try to send the data frame many times before the frame is successfully transmitted, which results in more power cunsumption of the mibile station.
Dynamic Backoff for Wireless Personal Networks by Ai-Chun Pang and Hsueh-Wen Tseng – NTU Taiwan – IEEE Globecom 2004.
Note : This resume is created for self-learning only. Author and Publisher hold copyrights
Reading – Listening – Talking – Writing – Programming – Mathematical Analysis
June 27, 2008
High Speed Network Lab