Age Control Protocol: Delivering Fresh Updates Over the Internet
Timeliness of information (state, sensor measurement, observation) available at interested recipients is key to enable cyber physical systems, for example V2X and public safety. We have sources that sense/ generate the information that is sent as update packets to recipients over a communication network. The generation process and network delays result in the information available at a recipient to be aged, that is the generation timestamp of information currently available at a recipient is older than the current time. We will focus on the workings of a recently proposed transport layer protocol, the Age Control Protocol (ACP), which enables timely delivery of update packets by sources to their recipients, in a network transparent manner, by appropriately adapting the rate of sending updates by the sources. To gain insight into age control, we will empirically compare ACP with a mix of loss-based, delay-based, and hybrid congestion control algorithms used by the Transport Control Protocol (TCP). While, as expected, state-of-the-art (hybrid) congestion control attempts to keep a number of bytes given by the product of the bottleneck rate and baseline round-trip-time (RTT) in the network pipe, the bottleneck rate and the baseline RTT may not shed as much light on the age optimizing rate of updates. To exemplify from our experiments, when TCP sends segments over an end-to-end path, consisting of an 802.11a 24 Mbps link followed by an intercontinental path over the Internet, it saturates the 802.11a link, which has the bottleneck rate for the path. Age, however, is optimized at a much lower rate than 24 Mbps. Turns out that the intercontinental path, much faster than the 802.11a link, is in fact the constraining factor with respect to the achievable age over the end-to-end path. We will also see that at the age optimal rate, depending on the network scenario, a source may send multiple updates per RTT or may send an update over many RTT. Age being optimized at low update rates has consequences with respect to multiple ACP end-to-end flows sharing a wireless access to send updates to recipients in the cloud. We will compare sharing of access amongst ACP flows with that amongst TCP flows. We will end the talk with some (half-baked) thoughts on future directions. [Work with Tanya Shreedhar at IIIT-Delhi and Roy D. Yates at Rutgers University.]
Sanjit Krishnan Kaul, IIIT Delhi
Sanjit K. Kaul is an Associate Professor in the Electronics and Communications Engineering department of IIIT-Delhi. He received the B.Tech. degree in electronics and communications engineering from the Birla Institute of Technology, Mesra, India, in 2000 and the Ph.D. degree in electrical and computer engineering from Rutgers University in 2011. He worked in the telecommunications industry for four years after completing his Bachelor’s degree. He was a Research Assistant with the Wireless Information Networks Laboratory from 2005 to 2011. He works on problems in wireless systems research. His current interests include age-of-information, networking for cyber-physical systems, and autonomous vehicles.