Network-Aware Adaptive Sampling for Low Bitrate Telehaptic Communication

Abstract

While the adaptive sampling technique for kinesthetic signal transmission offers a phenomenal reduction in the time-average data rate, it does not guarantee a meaningful upper bound on the instantaneous rate, which can occasionally be comparable to the peak rate. This implies that for Quality of Service (QoS) compliance, a network bandwidth equal to the peak rate must be reserved apriori for the telehaptic stream at all times. On a shared network with unknown and time-varying cross-traffic, this is not always feasible. In order to address the intermittently high bandwidth demand as well as the network-obliviousness of adaptive sampling, we propose NaPAS: Network-aware Packetization for Adaptive Sampling. The idea is to intelligently merge multiple haptic samples generated by adaptive sampling in a packet, depending on the changing network conditions. This results in an elastic telehaptic traffic that can adapt to the available network bandwidth. Through qualitative and quantitative measures, we evaluate the performance of NaPAS and demonstrate that it outperforms standard adaptive sampling (SAS) in terms of maintaining the haptic perceptual quality and QoS compliance, while also being friendlier to the exogenous network cross-traffic.