Issues on implementing new MAC protocol with Atheros 802.11x chips in Orbit Testbed

 

Zhibin Wu

 

Contents

1. About Orbit Testbed

2. MAC Experiments to run on Orbit Testbed

3. Current States of Open Source Linux Driver for Atheros 802.11 a/b/g Multi-band chip

4. What we need for the Atheros hardware?

 

1.      About Orbit Testbed.

 

The ORBIT (Open Access Research Testbed for Next-Generation Wireless Networks) system is a two-tier laboratory emulator/field trial network testbed designed to achieve reproducibility of experimentation, while also supporting evaluation of protocols and applications in real-world settings.  In particular, the laboratory-based wireless network emulator will be constructed using a novel approach involving a large two-dimensional grid of static and mobile 802.11a radio nodes which can be dynamically interconnected into specified topologies with reproducible wireless channel models.  All radio devices in the system provide open API¡¯s that permit end-users to download radio link, MAC and network layer protocols to construct a specific networking scenario.  Once the basic protocol or application concepts have been validated on the lab emulator platform, users can migrate their experiments to the field trial network which provides a configurable mix of both high-speed cellular (3G) and 802.11x wireless access in a real-world setting. 

 

2.  MAC Experiments to run on Orbit Testbed

 

The testbed is designed to support an open API for MAC protocol design. While the ultimate goal is to provide maximum flexibility for end-user¡¯s arbitrary MAC algorithm, we have a near-term goal to implement and verify a series of cutting-edge MAC protocols which is derived from IEEE 802.11 MAC protocol but has very important variations to improve the performance for traffics on wireless ad-hoc network. Following MAC experiments are going to be verified in Orbit testbed and all of them have some modifications of basic CSMA/CA mechanism.

 

1). DCMA (Data-driven Cut-through Multiple Access) Protocol. This new MAC protocol uses a new combo ACK/RTS frame to resolve the contentions within the same multi-hop flow. Thus, the frame of RTS and ACK needs to be changed. Correspondingly, it is necessary to implement a new control frame¡¯s handling routine for 802.11 chip¡¯s firmware or driver.

 

2) MACA-P protocol. This protocol uses an innovative way to establish parallel transmissions. Basically, the nodes which exchange an RTS/CTS reservation does not transmit the DATA frame immediately, but waiting for a time gap to let multiple transmission happen in parallel. Thus, both the format and timing relationship between the control frames and DATA frames should be adjusted.

 

3) D-LSMA protocol. Distributed Link Scheduling Multiple Access is design to facilitate real-time multi-hop flows. Using ns-2 simulation, it has been seen to give ~30% throughput improvement for wireless mesh network. The protocol suppresses the ACK frames and extends RTS/CTS format for support periodic traffic patterns. Thus, to realize this protocol, the ACK has to be disabled and the frame format needs to be changed.

 

3. Current States of Open Source Linux Driver for Atheros 802.11 a/b/g Multi-band chip.

   

While the MADWIFI (Multi-band Atheros Driver fro WIFI)  driver is under developing by "Sourceforge" and has released some versions of Linux driver, it has a HAL (Hardware Abstraction Layer) binary included which prohibits any direct handling of control frames and timing settings. The protocol state machine cannot be changed, except those related to management frames (such as beacon, association request).

Another available driver is "ar5K" developed by another group. The driver does list all the registers and has no obstruction to access them, but the driver program is not completely finished and most of wireless tools functionality left undeveloped. The driver itself is unstable. Based on current source code and a s very few embedded comments, we have no idea how to modify the control frame such as RTS/CTS and how to access and adjust the internal protocol-essential timers exactly.

 

4. What we need for the Atheros hardware.

 

l        Carrier sense scheme needs to be change such as sensing threshold. This is useful both for radio mapping purpose and MAC scheme

l        The timing relationship between the frames need to be adjust.

l        Need some changes in backoff scheme (such as timers settings)?

l        Be able to design new control frames.

l        Suppress ACK frames for unicast communications.

l        Need to change format of current control frames.

l        It is better to have bit error statistics from CRC function for channel estimation purpose.