Timely delivery versus bandwidth allocation for DASH-based video streaming over LTE

We study HTTP video streaming over a wireless access cell, such as LTE. We derive a closed form expression for the probability of timely delivery of the streamed video packets as a function of the bandwidth allocated to the user. The probability expression uses application, transport, and physical layer parameters relating to the employed video stream rate and to the quality of the communications channel. We validate the analytical results via numerical simulations based on the execution of real video traces and with LTE channel quality index trace modeling. The attained results allow the system designer, scheduler, and manager to calculate the bandwidth level that should be allocated to maintain acceptable system performance, expressed by two quality-of-experience (QoE) metrics, namely the video rate and the probability of timely delivery of video packets. The model provides a rationale and tools for performing admission procedures, bandwidth pricing policies, and cell dimensioning designs.