Scheduling URLLC and Punctured eMBB Users with Retransmissions and Rate Adaptation Techniques

In this paper, we consider a single Base-Station (BS) cellular system serving N u URLLC users and Ne eMBB users. Due to the stringent low latency requirements of the URLLC users, the scheduler needs to schedule their packets as soon as they arrive to the BS even if at the expense of a reduction in the eMBB user's rate. This is known in the literature as 'punc-turing'. To guarantee the ultra-reliability requirement of the URLLC users, the BS adapts the URLLC's rate according to the channel's state information. A low rate improves the reliability of the URLLC user at the expense of consuming more resource blocks which reduces the eMBB rate. To maximize the eMBB users' sum-rate while providing an acceptable URLLC reliability, we present an optimal scheduling algorithm and show its polynomial-time complexity in both Nu and Ne. We investigate this problem under two different retransmission schemes, namely, the Blind Retransmission and the Adaptive Retransmission. In addition, we show that our scheduling algorithms work even when the URLLC's channel state information is not available which is a typical case in more delay-stringent 5G applications.