Multiplex communications – Diagnostic testing – Determination of communication parameters
Reexamination Certificate
1999-12-15
2004-05-25
Olms, Douglas (Department: 2661)
Multiplex communications
Diagnostic testing
Determination of communication parameters
C370S232000, C706S015000, C706S016000
Reexamination Certificate
active
06741568
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention generally relates to characterizing communications between data processing systems and in particular to determining bottleneck link speed between data processing systems. Still more particularly, the present invention relates to employing adaptive resonance theory neural networks to determine bottleneck link speed between data processing systems with minimal sampling.
2. Description of the Related Art
Computer networks are essential features of contemporary computing, providing the framework for exchange of data and execution of distributed applications, either through client-server interaction such as HyperText Transmission Protocol clients and servers or collaborative operation such as failover redundancy in highly available networks. In contemporary networking environments, most customers have a need to control the use of existing bandwidth so that network management traffic is not competing with the business traffic.
Because some critical distribution tasks must be performed under certain constraints (e.g., download of new anti-virus software as quickly as possible or updating a data processing tool during non-business hours), knowledge of the speed of the physical link between different locations is critical to performing the tasks in an optimum manner. Knowledge of the bottleneck link speed, which represents the slowest link speed between two network entities, is therefore crucial in the sense that it will have a dramatic impact on the overall distribution process.
Currently there are two different techniques for obtaining the bottleneck link speed between two network entities. The first is to obtain a physical network topology, with all of the information on each of the links forming a part of the topology. The second is to compute the bottleneck link speed in an independent manner, preferably without the prerequisite of additional, expensive software. For transmission control protocol/Internet protocol (TCP/IP) environments, the most commonly employed method of independently computing bottleneck link speed is to send a sequence of ICMP ECHO packets from the source to the target and measure the inter-arrival times of the returning packets.
Because conditions are never ideal, and also due to the unstable nature of heterogenous networking environments, the independently measured bottleneck link speed value is never accurate when computed from only one sample. To obtain an accurate estimate of bottleneck link speed, multiple samples (typically hundreds, since accuracy is proportional to the number of samples collected) must be obtained and statistically processed. Different sized data packets transmitted at different intervals are employed, and the interval of the return data packets is compared to the transmission interval and analyzed—for example, through filtering and/or ordering or in a histogram—to determine the bottleneck link speed. However, these processes are extremely time consuming and computationally intensive.
It would be desirable, therefore, to provide a technique for quickly and accurately determining bottleneck link speed between two network entities. It would further be advantageous to be able to determine bottleneck link speed from a minimal number of data packet return interval measurements.
SUMMARY OF THE INVENTION
It is therefore one object of the present invention to provide an improved method, system and computer program product for characterizing communications between data processing systems.
It is another object of the present invention to provide an improved method, system and computer program product for determining bottleneck link speed between data processing systems.
It is yet another object of the present invention to employ adaptive resonance theory neural networks to determine bottleneck link speed between data processing systems with minimal sampling.
The foregoing objects are achieved as is now described. Bottleneck link speed, or the transmission speed of the slowest link within a path between two nodes, is determining by transmitting a sequence of ICMP ECHO data packets from the source node to the target node at a selected interval and measuring the return data packet intervals. Rather than using statistical analysis methods, the return data packet interval measurements are input into an adaptive resonance theory neural network trained with the expected interval for every known, existing network transmission speed. The neural network will then classify the return data packet interval measurements, indicating the bottleneck link speed. Since most of the computation—that required to train the neural network—may be performed before the data packet interval measurements are made rather than after, the bottleneck link speed may be determined from the return data packet interval measurements significantly faster and using less computational resources than with statistical analysis techniques. Moreover, fewer measurements are required to determine bottleneck link speed to the same degree of accuracy.
The above as well as additional objectives, features, and advantages of the present invention will become apparent in the following detailed written description.
REFERENCES:
patent: 5606664 (1997-02-01), Brown et al.
patent: 5684796 (1997-11-01), Abidi et al.
patent: 5710885 (1998-01-01), Bondi
patent: 5727157 (1998-03-01), Orr et al.
patent: 5751964 (1998-05-01), Ordanic et al.
patent: 5822535 (1998-10-01), Takase et al.
patent: 5845277 (1998-12-01), Pfeil et al.
patent: 5848243 (1998-12-01), Kulkarni et al.
patent: 5864862 (1999-01-01), Kriens et al.
patent: 6125105 (2000-09-01), Edwards
patent: 6209033 (2001-03-01), Datta
patent: 6215774 (2001-04-01), Knauerhase
patent: 6216163 (2001-04-01), Bharali
patent: 6260072 (2001-07-01), Rodriguez-Moral
The Incremental deployability of RTT-based congestion avoidance for high speed TCP Internet connections, Sigmetrics 2000, Santa Clara, Ca.*
Traffic Measurement and Analysis, kenjiro Cho, Sony 2002.*
U.S. patent application Ser. No. 09/465,181, Anstey et al., filed Dec. 15, 1999.
U.S. patent application Ser. No. 09/465,183, Barillaud, filed Dec. 15, 1999.
James W. Hong, Michael A. Bauer and J. Michael Bennett,The Role of Directory Services in Network Management, Sep. 28, 1992, pp. 175-187.
IBM Technical Disclosure Bulletin,Determination of Bandwidth and Latency in a Multimedia Communication Environment, Feb. 1994, vol. 37 No. 02A, pp. 349-350.
Joonho Park, Brian W. O'krafka, Stamatis Vassiliadis and Jose Delgado-Frias,Survey on Routers, Apr. 1994.
IBM Technical Disclosure Bulletin,Dynamic Determination of Network Topology, Mar. 1995, vol. 38 No. 03, pp. 411-418.
Chung Sheng Li, Yoram Ofek and Moti Yung,“Time-Driven Priority” Flow Control for Real-Time Heterogeneous Internetworking, Aug. 3, 1995.
Bracewell & Patterson L.L.P.
Dawkins Marilyn Smith
Olms Douglas
Pizarro Ricardo M.
LandOfFree
Use of adaptive resonance theory (ART) neural networks to... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Use of adaptive resonance theory (ART) neural networks to..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Use of adaptive resonance theory (ART) neural networks to... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3200014