Wireless Distributed Control Based on
The purpose of this work was to develop a testbed
for a new real-time wireless MAC protocol. This protocol is Robust Implicit
Earliest Deadline First (RI-EDF), an EDF-based algorithm for real-time wireless communication.
The intended use of this protocol is for distributed
real-time embedded systems like collaborative robots, or distributed control
applications where sensors/actuators are connected through a wireless
As testbed environment, we
decided to develop a wireless distributed control system
for an inverted pendulum. We created a demo in which we use camera sensors
in complement with potentiometer sensors on the cart to balance an
inverted pendulum. We used S.Ha.R.K.
for the real-time operating system on the computers. Berkeley
Mica2 motes running TinyOS were used
for wireless communication with RI-EDF protocol. We were able to successfully control the
pendulum and utilize the extra network bandwidth for other real-time communications on the same
shared channel using the proposed
This project was done in the Real-Time
Systems Lab of the Computer Science Department at the
University of Illinois at Urbana-Champaign.
website covers our development of the demo through the following 3
Distributed Control (based on RI-EDF): This is a
description of the final demo, using motes communicating by means of RI-EDF. Control of the
inverted pendulum was done using two switchable remote cameras for the
cart's position and one potentiometer on the cart for the rod angle.
Centralized Control: This controller
represents the first step we did for the project. Control is done by a
local machine using either both potentiometers on the cart or a camera
and a cart potentiometer. In both cases, the sensors were connected
directly to the controlling computer, so no network communication was involved.
Non-RT Wired Distributed Control: There were two intermediate
stages we passed through in moving from the centralized to the
distributed controller. We set up a remote and a distributed
controller over a non-real time 10-Megabit wired network using UDP,
instead of our wireless protocol. These served only to test that the
controller was working properly before adding RI-EDF. Notice that,
even if the network was non-real-time, the distributed control was able to work
properly because the 10-Megabit network was completely underloaded.
However, the real-time behavior became crucial when using MICA2 Motes
due to their limited bandwidth availability.
Overview of RI-EDF
Since distributed networks of wireless sensors/actuators can accomplish different real-time tasks ranging from space monitoring and surveillance to homeland security without human intervention, one focus of this research is to provide delay and throughput guarantee to real-time messages in wireless sensor networks. So far, we introduced a cellular-based architecture [Cac02, Cac03, Cac05a] suitable for wireless sensor networks and analyzed the network capacity when messages are scheduled with an implicit prioritized access protocol. More recently, the PI developed an asynchronous, robust real-time MAC protocol, named RI-EDF [Cre05, Cac05b, Cre06], which is suitable for real-time wireless communication among collaborative nodes or for distributed control in factory automation.
Wireless distributed sensing and control have the advantage that most of the real-time traffic is periodic and the wireless channel is inherently a multicast medium (whenever a node transmits, all nearby nodes receive the signal if collisions are avoided). These two properties are exploited to design a collision free MAC protocol based on EDF. The key idea is to replicate the EDF schedule at each node for packet transmission. If the schedules are kept identical, each node will know which one has the message with the shortest deadline and has the right to transmit next. Robustness and resource reclaiming are issues to be considered when developing a reservation based algorithm and all of them have been addressed when developing RI-EDF algorithm.
A testbed has been built which shows the applicability of RI-EDF for distributed control applications. RI-EDF has been extensively tested both by means of simulations (using ns-2) and by means of implementation on Berkeley MICA2 Motes.
M. Caccamo, L. Y. Zhang, L. Sha and G. Buttazzo,
"An Implicit Prioritized Access Protocol for Wireless Sensor Networks",
Proceedings of the IEEE Real-Time Systems Symposium, Austin, Texas, December 2002.
M. Caccamo and L. Y. Zhang, "The Capacity of Implicit EDF in Wireless Sensor Networks", IEEE Proceedings of the 15th Euromicro Conference on Real-Time Systems, Porto, Portugal, July 2003.
M. Caccamo and L. Y. Zhang, "The Capacity of an Implicit Prioritized Access Protocol in Wireless
Sensor Networks", Journal of Embedded Computing (JEC) by Cambridge International Science Publishing, Vol.1, No.2, 2005.
T.L. Crenshaw, A. Tirumala, S. Hoke, M. Caccamo,
"A Robust Implicit Access Protocol for Real-Time Wireless Collaboration",
Proceedings of the IEEE Euromicro Conference on Real-Time Systems, Palma de Mallorca, Spain, July 2005.
M. Caccamo, S. Hoke, T. Crenshaw, A. Tirumala, T. Luo, and J. Ha
"RI-EDF (Robust Implicit-EDF)",
Demo Session, ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc'05), Urbana-Champaign, IL, May 2005.
T.L. Crenshaw, S. Hoke, A. Tirumala, M. Caccamo,
"Robust Implicit EDF: A wireless MAC protocol for collaborative real-time systems",
To appear on ACM Transactions in Embedded Computing Systems (TECS), 2006.
Marco Caccamo (website,
T.L. Crenshaw (email)
This work is supported in part by the NSF
grants CCR-0237884 and CCR-0325716, ONR grant N00014-02-1-0102 and MURI