Distributed Cooperative Communication and Link Prediction in Cloud Robotics

Abstract

While deploying large scale heterogeneous robots in a wide geographical area, communicating among robots and robots with a central entity pose a major challenge due to robotic motion, distance and environmental constraints. In a cloud robotics scenario, communication challenges result in computational challenges as the computation is being performed at the cloud. Therefore fog nodes are introduced which shorten the distance between the robots and cloud and reduce the communication challenges. Fog nodes also reduce the computation challenges with extra compute power. However in the above scenario, maintaining continuous communication between the cloud and the robots either directly or via fog nodes is difficult. Therefore we propose a Distributed Cooperative Multi-robots Communication (DCMC) model where Robot to Robot (R2R), Robot to Fog (R2F) and Fog to Cloud (F2C) communications are being realized. Once the DCMC framework is formed, each robot establishes communication paths to maintain a consistent communication with the cloud. Further, due to mobility and environmental condition, maintaining link with a particular robot or a fog node becomes difficult. This requires pre-knowledge of the link quality such that appropriate R2R or R2F communication can be made possible. In a scenario where Global Positioning System (GPS) and continuous scanning of channels are not advisable due to energy or security constraints, we need an accurate link prediction mechanism. In this paper we propose a Collaborative Robotic based Link Prediction (CRLP) mechanism which predicts reliable communication and quantify link quality evolution in R2R and R2F communications without GPS and continuous channel scanning. We have validated our proposed schemes using joint Gazebo/Robot Operating System (ROS), MATLAB and Network Simulator (NS3) based simulations. Our schemes are efficient in terms of energy saving and accurate link prediction.