Swarm Intelligence for Routing in Communication Networks

1. Swarm Intelligence for Routing in Communication Networks Speaker: Shih-Chun Lin Advisor: Kwang-Cheng Chen

2. Outline • Introduction • Routing Algorithm • Swarm Intelligence • Swarm Intelligence Concept • Advantages • Example Algorithms • Swarm Intelligence Routing • Ant-Based Control (ABC) • AntNet • Multiple Round Trip Routing • Conclusion and Discussion • Reference 2

3. Outline • Introduction Swarm Intelligence Swarm Intelligence Concept Advantages Swarm Intelligence Routing Ant-Based Control (ABC) AntNet Multiple Round Trip Routing Conclusion and Discussion Reference 3

4. Introduction • Communication network managements is becoming increasingly difficult due to • Increasing network size • Rapidly changing topology • Complexity • The need for seamless interaction of numerous heterogeneous network components represents a formidable challenge, especially for networks that have traditionally used centralized methods of network control. This is true for both packet-switched and virtual-circuit networks. • Routing algorithm in modern networks • Performance metric • Average throughput and delay • Quality of service (Qos) guarantee • Classification • Centralized vs. distributed • Static vs. dynamic 4

5. Introduction • Swarm intelligence utilizes • Mobile software agents • Stigmergy (i.e. communication through the environment) Emergent behavior Simple interactions of autonomous, with simple primitives, give rise to a complex behavior 5

6. Outline Introduction • Routing Algorithm Swarm Intelligence Swarm Intelligence Concept Advantages Swarm Intelligence Routing Ant-Based Control (ABC) AntNet Multiple Round Trip Routing Conclusion and Discussion Reference 6

7. Routing Algorithm • Average throughput and delay • Centralized • Scalability • Inordinate demand for managing decisions requiring human attention • Inability of the network to recover in case of failure at the central controlling station • Static • Network conditions are time-invariant • Does not assess the traffic load • Adaptive • Inconsistencies arising from node failures • Potential oscillations that lead to circular paths and instability • Changes in the network occur too frequently to allow routing updates to propagate throughout all network nodes. • Combinatorially stable: if it changes sufficiently slowly for the routing updates to be propagated to all the nodes 7

8. Routing Algorithm • Minimal vs. non-minimal • Minimal: only minimal cost paths • Optimal routing • Shortest-path routing • Non-minimal: allows more flexibility in choosing the path by utilizing other heuristics • Qosguantee: pertaining to delay and bandwidth • Usually message based • Have the tendency to temporarily overuse network resources until they find the appropriate path 8

9. Outline Introduction • Swarm Intelligence • Swarm Intelligence Concept • Advantages • Example Algorithms Swarm Intelligence Routing Ant-Based Control (ABC) AntNet Multiple Round Trip Routing Conclusion and Discussion Reference 9

10. Swarm Intelligence Concept • Swarm intelligence gives rise to complex and often intelligent behavior through complex interaction of thousands of autonomous swarm members. • The main principle behind these interactions: Stigmergy or communication through the environment • Two examples for stigmergy • Pheromone laying on trails followed by ants • Autocatalytic reaction • Task-related stigmergy: alters the environment as to promote further similar action by the agents E.g. Sand grain laying by termites when constructing nests • Lay sand grains at random locations  A single heap reaches a critical mass, then construct their nest there 10

11. Advantages • Swarm intelligence boasts a number of advantages due to the use of mobile agents and stigmergy. • Scalability • Population and interactions of the agents • Fault tolerance • Graceful, scalable degradation other than catastrophic failure • Adaptation • Change, die, or reproduce • Speed • Changes in the network can be propagated very fast • Modularity • Agents act independently of other network layers • Autonomy • Little or no human supervision is required • Parallelism • Agent’s operations are inherently parallel 11

12. Example Algorithms [2] • Ant colony optimization (ACO): a class of optimization algorithms modeled on the actions of an ant colony • Useful in problems that need to find paths to goals • Artificial 'ants' - simulation agents - locate optimal solutions by moving through a parameter space representing all possible solutions. • Real ants lay down pheromones directing each other to resources while exploring their environment. • The simulated 'ants' similarly record their positions and the quality of their solutions, so that in later simulation iterations more ants locate better solutions. • Particle swarm optimization (PSO): a global optimization algorithm for dealing with problems in which a best solution can be represented as a point or surface in an n-dimensional space • Advantage: the large number of members that make up the particle swarm make the technique impressively resilient to the problem of local minima • Stochastic diffusion search (SDS): an agent-based probabilistic global search and optimization technique best suited to problems where the objective function can be decomposed into multiple independent partial-functions 12

13. Outline Introduction Swarm Intelligence Swarm Intelligence Concept Advantages • Swarm Intelligence Routing • Ant-Based Control (ABC) • AntNet • Multiple Round Trip Routing Conclusion and Discussion Reference 13

14. Ant-Based Control (ABC) [3] • The algorithm is adaptive and exhibits robustness under various network conditions. • It’s accomplished by using agents, called ants, who deposit virtual pheromone on the trails they follow, which affects the routing tables of the affected nodes. • The probabilities of the routing tables are updated as the ant visits the nodes. • Life time of the ant at the time of the visit (T): • Delay (Di) : • Step size for the visited node (δr): • The routing table is then updated according to: 14

15. Ant-Based Control (ABC) [3] • The ant both uses and updates the routing table at the same time. • E.g. if the source is node F and the destination is node E, then the ant will update the row for F and use the node for E to find the next hop • Update rules: [1] • Cooperation of an exploration factor g • With probability (1-g) the ants are forwarded with a total uniform distribution; with probability gthe ants are forwarded according to the routing table probabilities 15

16. AntNet [4] • Routing is determined by means of very complex interactions of forward and backward network exploration agents (“ants”). • The idea behind this subdivision of the agents: Allow the backward ants to utilize the useful information gathered by the forward ants on their trip from source to destination. • The entries of the routing table are probabilities, and as such, must sum to 1 for each row of the network. • These probabilities serve a dual purpose: • The exploration agents of the network use them to decide the next hop to a destination, randomly selecting among all candidates based on the routing table probabilities for a specific destination • The data packets deterministically select the path with the highest probability for the next hop [1] 16

17. AntNet [4] • The sequence of actions is simple and intuitive: • Each network node launches forward ants to all destinations in regular time intervals. • The ant finds a path to the destination randomly based on the current routing tables. • The forward ant creates a stack, pushing in trip times for every node as that node is reached. • When the destination is reached, the backward ant inherits the stack. • The backward ant pops the stack entries and follows the path in reverse. • The node tables of each visited node are updated based on the trip times. [1] 17

18. AntNet [4] • Except for the routing table, each node also possesses a table with records of the mean and variance of the trip to every destination. [1] • From the trip time table to the routing table: • Intermediate quantity in the processing of the raw trip time information (r’): The ratio of the variance to the mean (σ /μ) is used as a measure of the consistency of the trip times. 18

19. AntNet [4] With the case where the consistence is high and the time is good, we want the processed r’ to be even smaller. • Further positive or negative reinforcement of good or bad routes takes place next, via negative feedback. [1] • Update the routing table probabilities. • The packets of the network then use these probabilities in a deterministic way, choosing as next hop the one with the highest probability. 19

20. Multiple Round Trip Routing [5] • The basic version utilizes the cost measured by the forward ants to update the routing table entries. • As in AntNet, nodes launch forward ants in regular intervals. • The forward ant keeps track of the visited nodes in a stack Jk and of their associated cost dkn,d. • Once the destination d is reached, then a backward ant is launched, which updates the distance estimation for node n to d via j as follows:  Routing table probabilities are updated via: 20

21. Multiple Round Trip Routing [5] • The interesting improvement is based on Bellman’s principle of dynamic programming. • Every node in the path Jkof a source-destination pair s-d, is considered a destination. [1] 21

22. Outline Introduction Swarm Intelligence Swarm Intelligence Concept Advantages Swarm Intelligence Routing Ant-Based Control (ABC) AntNet Multiple Round Trip Routing • Conclusion and Discussion Reference 22

23. Conclusion and Discussion • AntNet and Multiple Trip Routing are two examples of the class of swarm intelligence algorithms that incorporate round-trip agents. • In this type of algorithms, the forward ants act as investigators and the backward ants are the ones who update the routing tables. • ABC is an algorithm that incorporates only forward agents, who perform the update as they travel through the network. • In this type of algorithms update is faster and more reliable, since there is no delay between the information gathering and the actual update. 23

24. Outline Introduction Swarm Intelligence Swarm Intelligence Concept Advantages Swarm Intelligence Routing Ant-Based Control (ABC) AntNet Multiple Round Trip Routing Conclusion and Discussion • Reference 24

25. Reference • I. Kassabalidis, M. A. El-Sharkawi, R. J. Marks II, P. Arabshahi, and A. A. Gray, “Swarm intelligence for routing in communication networks,” Global Telecommunications Conference, 2001. GLOBECOM ‘01. IEEE , vol.6, no., pp.3613-3617 vol.6, 2001. • http://en.wikipedia.org/wiki/Swarm_intelligence • R. Schoonderwoerd, O.E. Holland, J. Bruten, and L. Rothkrantz, “Ant-based load balancing in telecommunications networks,” HP Labs Technical Report, HPL-96-76,May 21, 1996. • G. Di Caro and M. Dorigo, “AntNet: a mobile agents approach to adaptive routing ,” Tech. Rep. IRIDIA/97-12, Université Libre de Bruxelles, Belgium. • M. Heusse, D. Snyers, S. Guérin, and P. Kuntz, “Adaptive agent-driven routing and load balancing in communication network,” Proc. ANTS'98, First International Workshop on Ant Colony Optimization, Brussels, Belgium, October 15-16, 1998. 25