On tile assignment for region-of-interest video streaming in a wireless LAN

Abstract

We consider the following problem in this paper: A video is encoded as a set of tiles T and is streamed to multiple users via a onehop wireless LAN. Each user selects a region-of-interest (RoI), represented as a subset of T , in the video to watch. The RoI selected by the users may overlap. Each tile may be multicast or unicast. We define the tile assignment problem as: which subset of tiles should be multicast such that every user receives, within a transmission deadline, the subset of tiles pertaining to the RoI the user selected, while minimizing the number of unwanted tiles received by users. We present and evaluate five tile assignment methods. We show that: (i) minimizing transmission delay can lead to significant wasteful reception in the multicast group, (ii) using tile access probability to assign tiles frequently leads to assignments that violate the deadline, and (iii) a fast, greedy, heuristic works well: it performs close to the optimal method and can always find an assignment within the deadline (as long as such assignment exists).

Tiling of panorama video for interactive virtual cameras: Overheads and potential bandwidth requirement reduction

“…In this respect, researchers in the multimedia community have for some time analyzed region-of-interest streaming solutions. For example, tiling is discussed in [6], [7], [8], [9], [10]. Furthermore, [11], [12], [13], [14], [15] extensively address the problem of automatically generating personalized content, and [16] discusses plain cropping.…”

“…The works related to tiling can be broadly classified into two scenarios, the server side generation [6], [8], [9], [10] and the client side generation [17], [7]. The former emphasize on reducing the through-put on the server side because their processing happens on server and the video is merely transferred to the client for display.…”

Predicting head trajectories in 360° virtual reality videos

“…In [1], bandwith adaptive streaming techniques are proposed by making tiles of different size and quality available and optimizing the selection of streams depending on the network parameters. Several bandwidth usage optimization algorithms are proposed in [8] where multiple users are sharing a wireless network in a region of interest (RoI) video streaming scenario. Based on utility functions considering network conditions and overlapping user RoIs, the algorithms decide on which tiles are to be streamed and whether to use unicast or multicast.…”

“…The domain specific predictor works with several rules designed to keep the tracked ball and players near the RoI centre based on their distances to RoI centre and intensities in the residual from the previous frame. In the previously mentioned [8], the authors tried modifying the utility functions to use the access statistics of the tiles from previous viewings to help the tile selection process, however the wrong predictions based on the probabilistic access patterns were judged to cause too many invalid tile assignments. In [2] and [19] models for view centre prediction are proposed which utilize linear regression of a short window of previous view centres to estimate future view centres.…”

Interactive Zoom and Panning from Live Panoramic Video

“…If processed on the server side, we support a limited number of users per machine. However, taking into account existing work to transfer panorama video in realtime [6] and the fact that GPUs are becoming a commodity also on mobile phones, we work on an adaptation of our system that creates the virtual camera on the client side. This system will trade bandwidth for server processing power by broadcasting the panorama video to all clients instead of computing and transmitting a personalized view for each client.…”