Abstract
Aloha is a random access protocol that was developed at the University of Hawaii to connect geographically separated computer systems via a radio network. The different transmitters transmit their information, which comes in blocks (packets) of fixed size and duration T, whenever they have something to send. This causes possible collisions of packets if two stations send at the same time. If collisions occur, the transmitted packets are annihilated and put into a random delay buffer to be sent again some time later. Therefore, in this project we propose to study the effect of various factors on the system performance such as, the number of receivers, number of retries, mean connection duration, the mean back-off time, and the number of codes assigned to access the channel in order to increase the throughput or efficiency of Aloha. Currently, the throughput of Aloha sits at around 18.6%. We hope to dramatically increase the efficiency of this random access method through our research and work.
II. Project Proposal Plan
II-1. Introduction
Due to the recent failure of communication systems in emergency situations, similar to the Hurricane in New Orleans [1], improving ad hoc wireless networks has become a necessity for disaster relief agencies. The main goal of our project is to improve the efficiency of ad hoc wireless networks through Medium Access Control (MAC) protocol design.
Essentially all ad hoc wireless networks can be described in terms of the Open Systems Interconnection (OSI) Reference Model. The OSI Reference Model splits the different processes, which data in a wireless system follows, into seven layers. Our analysis will involve the bottom three levels, network, data and physical, which basically controls how the data is sent between transmitter and receiver [2]. Aloha protocol, which is crude and inefficient, will be used as the basis for the MAC protocol in our design analysis. Inherently, the efficiency of Aloha protocol is about 18%.
By varying different parameters, the efficiency of the Aloha throughput can be increased. For example Slotted Aloha, which is a slight modification of Aloha (Pure Aloha, see A1), has almost double the throughput efficiency [3]. It was shown in [4] that by taking into account the state of the channel used for transmission, rather than assuming them identical, the efficiency of throughput for slotted aloha can also be increased. A different method proposed by [5], involving enforcing a pricing mechanism on the user transmitting data, also shows improved performance. Another approach, which was also able to increase the efficiency, introduced by [6], involves using “smart”
antennas and Direction-Of-Arrival (DOA) algorithms to predict the tendency of the received signal.
II-2. Design Requirements
To increase the efficiency of an Aloha ad hoc wireless system, the group will continue along the lines of [4], [5] and [6] by adjusting different parameters, such as the number of receivers, number of retries, mean connection duration, the mean back-off time, the number of codes assigned to access the channel, etc., in a computer based simulation.
Ideally the group would like to achieve a throughput of close to 50% efficiency, which would be an improvement on the Slotted Aloha maximum throughput of 36.8%. However, any increase in efficiency would be desirable.
II-3. Design Approaches
In designing a system to increase the efficiency of the Aloha protocol, two main approaches will be employed: simulation via software and simulation via hardware. Initially, it is necessary to develop a computer-based simulation of the protocol in order to see how adjustments to various parameters yield changes in efficiency. A computer-based simulation will allow the group to quickly track how changes to various parameters (including the number of receivers, number of retries, mean connection duration, the mean back-off time, and the number of codes assigned to access the channel) effect overall system efficiency. Once a computer-based simulation has produced a system of desirable efficiency, it will be up to the group to verify these results using actual hardware.
The computer simulation will ultimately be made using either Matlab/Simulink or Opnet. Matlab and Simulink, however, are the best choice for initial programming, since the group is most familiar with these programs. Simulink offers the advantage of being graphically-based, making it easier for members of the group to isolate and explain various parts of the simulation to one-another. Additionally, Simulink's interface allows
the user to quickly make changes to the system and to track the results of such changes. Opnet was suggested by Professor Tureli as a possible alternative to Matlab and Simulink. The advantage of Opnet is that it is designed specifically to model communications systems. For the time being, Simulink and Matlab appear to be sufficient, though it may become necessary to use the more specialized Opnet. Should Opnet be needed, it will be necessary for some or all of the members of the group to master the program. Regardless of which software is used, the final design approach will be determined using one main criterion: efficiency. That is, the system with the highest efficiency (# packets received / # packets transmitted) will ultimately be the most desirable, regardless of which software is used to do the simulation.
The main disadvantage of the computer-based simulation is that it will necessarily be based on some set of assumptions, and being such it will always fall short of the actual system. For example, the simulations are typically based on an unlimited number of users, whereas in reality there will be a fixed number of users. As a result, it is necessary for the group, after completing the computer simulation, to implement the simulation on actual hardware. Although this will once again be a simulation that is not necessarily exactly the same as the true system, it should be closer than the software simulation. In addition to verifying or disproving the results of the computer simulation, the hardware simulation will provide a nice demonstration to display on design day, and may help to highlight some of the shortcomings of a software-based design. The hardware to be used will be provided by Professor Tureli, as he has a well-equipped wireless communications
laboratory. Specific devices have not yet been considered, because the group is still at the stage of implementing the simulation solely on software.
II-4. Financial Budget
|
Labor Cost: |
|
|
|
|
|
|
# of Engineers |
Price per hour per person |
Hours per week |
# of weeks |
Cost per Engineer |
Total |
|
4 |
$25.00 |
4 |
28 |
$2,800.00 |
$11,200.00 |
The Budget numbers, on the previous page, are an estimate of what the materials costs. Some of these items might be available to borrow from another group or university department. These are only estimates, and might see significant revisal during the semester as more details become available.
II-5. Project Schedule
|
Task Description |
Days |
Member Initials |
Start |
Finish |
|
|
|
|
|
|
|
Research |
14 |
|
9/4/05 |
9/19/05 |
|
Possible Research Funding |
|
MT KD CB DT |
|
|
|
Research of Technology |
|
CB DT |
|
|
|
Research of Simulation Technique |
|
KD DT |
|
|
|
Problem Definition |
21 |
|
9/19/05 |
10/10/05 |
|
Technical Research |
|
MT KD CB DT |
|
|
|
Market Research |
|
KD DT |
|
|
|
Design Process |
35 |
|
10/10/05 |
11/15/05 |
|
Identifying Design Requirements |
|
CB DT |
|
|
|
Design Documentation |
|
MT KD |
|
|
|
Testing |
35 |
|
11/15/05 |
2/20/05 |
|
Actual Testing |
|
MT KD |
|
|
|
Analysis of Results |
|
CB DT |
|
|
|
Further Testing |
|
MT DT |
|
|
|
End Product Documentation |
38 |
|
2/20/05 |
3/30/05 |
|
Final Report |
|
MT KD CB DT |
|
|
|
Final Presentation |
|
MT KD CB DT |
|
|
|
Website |
|
MT KD |
|
|
III. Conclusion
Efficient communication systems are of vital importance in today's world. In times of emergency and tragedy, there is a great number of information being transmitted. Due to this great load collisions begin to occur and, eventually, the communication system fails. Aloha has been around for many years. However, its efficiency is quite poor and sits at approximately 18%. We chose to work with Aloha for two main reasons: (1) it does not require channel availability measurement by the users and (2) it does not require any timing coordination between the different users. This is particularly suitable because in a world where a lot of information is being transmitted on a daily basis it is practically impossible to synchronize the users due to the large delays. Therefore, we propose to increase the throughput of Aloha through experimentation with various factors including connection time and number of receivers. First, we propose to create a code in either Matlab or Simulink. Then, we will implement this code on hardware possibly provided to us by Prof. Tureli and we will see the effects of experimenting with certain factors, such as those mentioned in this report. We hope to greatly increase the efficiency of Aloha; hopefully, to about 50%. We fully acknowledge that a lot of hard work lies ahead but we expect that our design approaches will yield a successful result at the conclusion of our senior design project.
REFERENCES
[1] Wikipedia. Effect of Hurricane Katrina on New Orleans.
http://en.wikipedia.org/wiki/Effect_of_Hurricane_Katrina_on_New_Orleans
[2] J. Tyson, How OSI Works. How Stuff Works. http://computer.howstuffworks.com/osi.htm
[3] ALOHA Protocol. Lay Networks. http://www.laynetworks.com/ALOHA%20PROTOCOL.htm
[4] S. Adireddy and L. Tong. Medium Access Control with Channel State Information for Large Sensor Networks. MMSP-2002. St. Thomas, U.S., December 11, 2002.
[5] D. Wang, C. Comaniciu and U. Tureli. Cooperation and Fairness for Slotted Aloha. Department of Electrical and Computer Engineering at Stevens Institute of Technology. September 30, 2005.
ALOHA: SIMULATION AND CRITIQUE OF A RANDOM ACCESS METHOD
Group #11
Date Due:
October 18, 2005
Faculty Technical Advisor:
Prof. Uf Tureli ______________________________
Group Members:
Christian Budiarjo ___________________________
Kristen Daly________________________________
Michelle Teixeira____________________________
David Toner________________________________
"I pledge my honor that I have abided by the Stevens Honor Code."
