Exploiting Chemistry for Better Packet Flow Management 5: Chemical Congestion Control, Design Motifs, and Conclusion

# Exploiting Chemistry for Better Packet Flow Management 5: Chemical Congestion Control, Design Motifs, and Conclusion

This represents the final installment of the series reviewing the 2011 technical report by Meyer and Tschudin. Part 1 gave an overview of the report and the problems it aimed to solve, as well as the chemistry basics necessary for further understanding. Part 2 discussed artificial chemistries and the extension to an artificial packet chemistry, setting the mathematical framework for analyses and implementation. Part 3 discussed briefly some methods of network analysis available once the artificial packet chemistry was developed. Part 4 gave an implementation of a scheduler based on the packet chemistry model as well as the idea of a chemical control plane and its implementation. This final part will discuss one final application — a congestion control algorithm–as well as mention design motifs pointed out by the authors and conclude our analysis of this report.

## Chemical Congestion Control Algorithm

As a final application of chemical networks, the authors combine LoMA-scheduled queues and flow-filtering patterns to schedule segments of a transport protocol to control congestion. To illustrate, they re-implement the additive increase/multiplicative decrease of the congestion avoidance mode of TCP-Reno. The congestion control algorithm reacts to packet loss automatically and naturally.

The reproduced figure above shows how the chemical congestion control is implemented as a chemical reaction network (and by extension, a queueing network).

1. Arriving packets are put into a queue $D$. The transmission rate $\nu_{tx}$ is controlled  by the quantity of pacemaker molecules $R$, so $\nu_{tx} = k_{1}c_{R}c_{D}$, once again according to the Law of Mass Action. To mimic the additive (linear) increase mechanism of TCP-Reno, the number of pace-maker molecules is increased at a rate $\nu_{\text{inc}}$.
2.  Before packets are transmitted, they are tagged with a sequence number. If there is a gap in the sequence number of acknowledgments from the destination, the source regenerates the packets at a queue $L$.
3. A lost packet will catalyze the destruction of pacemaker molecules by another reaction $r_{2}$, which will lead to the exponential decay of $R-$molecules and thus decrease the transmission rate. However, we wish to prevent too fast a destruction of pacemaker molecules, so a third reaction $r_{3}$ will delay the destruction.

The authors encourage the use of such a reaction graph at the design phase of flow management policies. The feedback nature is much clearer. In addition, the papers give a formal proof that their congestion control model is TCP-fair at equilibrium; that is, the transmission rate is proportional to $\frac{1}{\sqrt{p_{\text{loss}}}}$ where $p_{\text{loss}}$ is the probability of packet loss between source and destination. They also discuss an extended version that reacts to variations in round trip time (RTT) variation to more fully exploit the link bandwidths. The traffic statistics are not computed symbolically with chemical reactions. Instead, another reaction builds a difference sch that at equilibrium the fill level of its queue is proportional to the excess transmission rate. That signal decays the pacemaker molecules. They also supply simulations to illustrate their implementations.

## Design Motifs

This section is of a particular interest to me personally, and was treated the least. No design process can ever be fully automated, but the authors claim to have developed several design motifs of chemical reaction networks for a variety of purposes, including arithmetic computation of fill levels to communication patterns (anycast, neighborhood discovery, etc). Unfortunately, they do not give a direct citation as to where to look further. This report will be updated when such information is found.

Figure 12 shows the only two motifs provided by the authors. One for rate limiting (a) and the other bimolecular reaction to compute the difference between arrival rates for two queues. The concept of using these as design elements is extremely intriguing, and it was unfortunate that the authors did not choose to expand this further.

## Conclusion

Meyer and Tschudin have given an extensive report showing how powerful the application of chemical kinetics and chemical networks can be for the computer networking space. There are several research opportunities available for further study and implementation. As yet, there have been no citations of this work of note (the report came out in 2011), and thus the opportunity seems ripe for exploration.

## References

1. Dittrich, P., Ziegler, J., and Banzhaf, W. Artificial chemistries – a review. Artificial Life 7(2001), 225–275.
1. Feinburg, M. Complex balancing in general kinetic systems. Archive for Rational Mechanics and Analysis 49 (1972).
2. Gadgil, C., Lee, C., and Othmer, H. A stochastic analysis of first-order reaction networks. Bulletin of Mathematical Biology 67 (2005), 901–946.
3. Gibson, M., and Bruck, J. Effcient stochastic simulation of chemical systems with many species and many channels. Journal of Physical Chemistry 104 (2000), 1876–1889.
4. Gillespie, D. The chemical langevin equation. Journal of Chemical Physics 113 (2000).
5. Gillespie, D. The chemical langevin and fokker-planck equations for the reversible isomerizationreaction. Journal of Physical Chemistry 106 (2002), 5063–5071.
6. Horn, F. On a connexion between stability and graphs in chemical kinetics. Proceedings of the RoyalSociety of London 334 (1973), 299–330.
7. Kamimura, K., Hoshino, H., and Shishikui, Y. Constant delay queuing for jitter-sensitive iptvdistribution on home network. IEEE Global Telecommunications Conference (2008).
8. Laidler, K. Chemical Kinetics. McGraw-Hill, 1950.
9. McQuarrie, D. Stochastic approach to chemical kinetics. Journal of Applied Probability 4 (1967), 413–478.
10.  Meyer, T., and Tschudin, C. Flow management in packet networks through interacting queues and law-of-mass-action-scheduling. Technical report, University of Basel.
11.  Pocher, H. L., Leung, V., and Gilles, D. An application- and management-based approach to atm scheduling. Telecommunication Systems 12 (1999), 103–122.
12. Tschudin, C. Fraglets- a metabolistic execution model for communication protocols. Proceedings of the 2nd annual symposium on autonomous intelligent networks and systems (2003).