Directional antenna ad hoc network neighbor discovery method based on SARSA (lambda) algorithm

Abstract

The invention discloses a directional antenna ad hoc network neighbor discovery method based on an SARSA (lambda) algorithm, and the method comprises the following steps: 1, enabling each node to initialize an own Q matrix at the beginning of a neighbor discovery process, and randomly selecting an initial state and an initial action; 2, enabling the node to enter a transmission mode and a receiving mode to carry out corresponding steps, and adopting different steps; 3, employing a greedy strategy to select the action of the next time slot; 4, calculating a one-step prediction error; 5, for all states and actions, updating a Q matrix according to a one-step prediction error; 6, entering the next time slot, returning to the step 2, and stopping until the neighbor discovery process is finished. Compared with a complete random algorithm, the ad hoc network neighbor discovery method disclosed by the invention can accumulate experience in scanning and adaptively find sectors with undiscovered neighbors, so the neighbor discovery speed is accelerated.

Description

technical field [0001] The invention belongs to the field of wireless network communication, in particular to a method for discovering neighbors in a directional antenna ad hoc network based on a SARSA (λ) algorithm in the field. Background technique [0002] The application scenarios of wireless ad hoc networks are becoming more and more complex, and directional antennas are used in them due to their advantages of long transmission distance and strong anti-interference ability. However, due to its limited beamwidth, it brings great challenges to the neighbor discovery process in the network. The SARSA(λ) algorithm can continuously accumulate experience to update the strategy according to the information scanned in the network in the past, so that it gradually converges to the optimal strategy, thus speeding up the speed of neighbor discovery. [0003] Most of the current neighbor discovery algorithms use omnidirectional antennas in the whole process or in part of the proce...

AI Technical Summary

Problems solved by technology

Among them, it is mainly divided into random algorithm and planning algorithm. The random algorithm is inefficient and is greatly affected by different network structures. The planning algorithm needs to be coordinated by a central node, which will make the network more complicated.

Some algorithms add GPS and other location aids to improve efficiency, but it will increase the power consumption of nodes, which is not suitable for wireless ad hoc networks

Images