A model of Fuzzy Control Backoff Schemes in Telecommunication Networks SOMCHAI LEKCHAROEN 1 AND CHANINTORN JITTAWIRIYANUKOON 2 1 Faculty of Information Technology, Rangsit University THAILAND 2 Faculty of Science and Technology, Assumption University THAILAND Abstract:- Fuzzy control is based on fuzzy logic, which provides an efficient method to handle inexact information as a basis of reasoning. With fuzzy logic it is possible to convert knowledge, which is expressed in an uncertain form, to an exact algorithm. In fuzzy control, the controller can be represented with if-then rules. The interpretation of the controller is fuzzy but the controller is processing exact input-data and is producing exact output-data in a deterministic way. However, Backoff time computation schemes, namely: pseudorandom backoff (PB) time, exponential backoff (EB) time and random backoff (RB) time that are applicable in waiting time re-arrangement in queue. They have proved to be inefficient in coping with the conflicting requirements, that is, low dropping frames and high conforming frames. This led us to explore alternative solutions based on artificial intelligence techniques, specially, in the field of fuzzy logic. In this paper, we propose a fuzzy control backoff scheme that aims at detecting violations of parameter negotiated. We evaluate and compare the performance of fuzzy control backoff scheme (FB) with namely, pseudorandom backoff scheme (PB), random backoff scheme (RB) and exponential backoff scheme (EB). The performance of four backoff schemes have been investigated by the fluctuation of telecommunication traffic stream (burst/silent type). Simulation results show that the fuzzy control scheme helps improve performance of our re-arrangement waiting time in queue compared to other non-fuzzy backoff schemes. Keywords: Backoff time computation, fuzzy control, policing mechanisms. 1 Introduction Backoff computation, each source will delay the message whenever the transmission to next service fails. Backoff algorithms have introduced many techniques such as exponential backoff, random backoff, linear backoff and quadratic backoff as described in many papers [1],[2],[3]. For example, messages sent by senders in an Ethernet network may be retransmitted after T steps where T is selected randomly from {1,2,3, . . . ,2 min(10,b) } and b is the number of times the station has tried to send the packet but failed. This is one of an example of general application referred to exponential backoff. In this paper, we apply backoff concepts to waiting time in the queue with policing mechanisms and evaluate the performance using a high speed network model. 1.1 Backoff algorithm Many papers study the backoff algorithms in terms of their effect on network performance as the offered load increases. However, simplification or modification of backoff algorithm can lead to very different analytical results [3],[4]. Many backoff schemes have been proposed and studied. 1.1.1 Pseudorandom backoff In pseudorandom backoff (PB) scheme, none of the computation is applicable but queue disciplines. They are FIFO, LIFO and priority. In this paper, the FIFO and the maximum queue size are preset. 1.1.2 Exponential backoff Exponential backoff (EB) is an algorithm being widely used in traffic offered load. In EB, each node doubles the backoff time after each retry occurs (2x) but not above the maximum value (B max ), and decreases the backoff interval to the minimum value (B min ) after a successful retry. We summarize EB by the following set of equations: x ä min(2x,B max ) upon retry and x ä B min upon successful transmission. The x is the backoff interval value. The values of the B max and B min are predetermined, based on the possible range of number of active nodes and the traffic load of a network. For example, B max and B min are usually set to 1024 and 2, respectively. Although some researchers found that the channel throughput in the Ethernet network will be degraded as the backoff interval does not correctly represent the actual contention of the channel [5],[6],[7] but we experience somehow the EB can help improve the performance of the system regarding to the fluctuation of telecommunication traffic. 4th WSEAS Int. Conf. on COMPUTATIONAL INTELLIGENCE, MAN-MACHINE SYSTEMS and CYBERNETICS Miami, Florida, USA, November 17-19, 2005 (pp24-28)