A cooperative protocol for video streaming in dense small cell wireless relay networks

https://doi.org/10.1016/j.image.2014.12.005Get rights and content

Highlights

  • We develop a protocol for using overheard video packets in wireless relay networks.

  • We develop a wireless relay selection scheme that uses the content of the video packet and not just the channel quality.

  • Overhearing is enough for improving wireless unicast video quality when compared to network coding.

  • Video-aware relay selection outperforms classic channel-quality based relay selection.

Abstract

The small-cell wireless network (SCN) paradigm is based on the deployment of low-power small cells close to the user so that spatial re-use is increased. Small cell base stations (SCBSs) can be connected either with a wired or a wireless backhaul link to the core network. However, since these SCBSs are expected to be deployed in significant numbers [1], the increased density creates not only new problems but also optimization opportunities. In this paper we work towards the full exploitation of the broadcast advantage that dense SCNs offer in order to improve the video quality of unicast streaming applications.

To accomplish our goal we design a protocol that exploits the broadcast channel in the neighborhood of each small cell by using the SCBSs opportunistically as a relay stations. Each SCBS can overhear packet transmissions both from the macro BS (MBS) and also from other SCBSs in its neighborhood. Furthermore, since a packet may be available at multiple SCBSs, we propose an algorithm for optimized relay selection that takes into account the content of each specific video packet. This algorithm selects jointly the optimal small cell relay and video packet for forwarding. Both the overhearing protocol and the relay selection algorithm use only passively collected information and require no explicit message passing. The evaluation of the two proposed sub-systems shows significant performance improvements under different key system configurations for the video content, the number of neighboring SCBS, and the type of the backhaul connection.

Introduction

High quality video streaming in wireless networks is one of the most popular mobile applications today. The widespread adoption of mobile devices that are capable of handling sophisticated video processing and high data-rate wireless communication algorithms is propelling this demand. The video traffic explosion in wireless networks is expected to accelerate even more the next few years [2]. To ensure high quality video streaming there is a need for more bandwidth. Even after compression, video is bandwidth-hungry and delay-sensitive. Thus, there is no fundamental way to bypass this problem except by providing more bandwidth. To ameliorate this problem in cellular networks, small cell base stations (SCBS) or small cell relay stations are deployed closer to the user in order to increase network capacity and spatial coverage respectively.1 The envisioned cellular network configuration includes several SCBS with overlapping coverage, while the typical macro base station (MBS) provides umbrella coverage [1], [3]. Thus, besides the single SCBS node that a user may be associated to, neighboring SCBSs will also probably be reachable by the same user (see our topology in Fig. 1). But the dense deployment of these wireless networks can be exploited with even smarter ways so that they can increase the capacity, decrease delay, and minimize susceptibility to channel variations if they operate as relays. Although cooperative diversity with relays has been investigated considerably from a theoretical perspective, in the immediate future the prospects of being implemented are better than ever precisely because of the high demand for increased bandwidth and coverage. Many standards like LTE-A support relay-based transmission modes that have been shown to be practical [4].

In this paragraph we will try to highlight potential optimization opportunities in dense SCNs with simple examples. Consider the case of a wireless backhaul connection between the MBS and the SCBSs as illustrated in Fig. 1. The transmissions from the MBS can be overheard from the two target SCBSs within the cell. Now the question that has to be answered is which of the two SCBSs will transmit a packet from the first flow since they have both received it. Also assume that SCBS B1 has been granted access to the channel and transmits the packet (according to one of the algorithms that we will propose). Due to the proximity of the two SCBSs, SCBS B2 cannot transmit simultaneously with SCBS B1. However, this does not prevent it from overhearing the transmission of packets from its neighbor, and remove it from its local buffer when user D1 acknowledges it (this is handled by the second algorithm we propose in this paper). The same situation will occur if more SCBSs are deployed close to SCBS B1, i.e., they can also overhear. When the SCBSs connect to the core network through a wireline backhaul, the network topology is similar only in this case SCBS B2 can only overhear a packet transmission from SCBS B1 and not the MBS. From the previous scenarios, we see that in dense SCNs there will be ample opportunities for collecting information from the neighborhood of each SCBS. In this paper we aim to exploit to the fullest these opportunities that arise from a fundamental property of a wireless channel which is its broadcast nature.

The first central idea of our proposal is to allow SCBSs to passively overhear data packets, so that several nodes have available the same information (packet). This means that the SCBS, who has the opportunity to transmit a packet, can do so by selecting from a higher number of candidate packets. The second central idea of this paper is to leverage these overheard packets that are available in the multiple SCBSs in order to obtain a diversity gain for the transmitted signals, but in a way that is aware of the video content of the packet that will be transmitted.

The previous ideas are implemented with two protocols/algorithms. Our concrete contributions that build incrementally one on top the other are

  • 1.

    We propose a lightweight packet overhearing protocol for dense SCNs that operates between the SCBSs and is based on the concept of opportunistic communication. The cooperative protocol allows the SCBSs to overhear packets and ensures that duplicate video packet forwarding from different SCBSs is avoided.

  • 2.

    After the previous protocol is applied and ensures that no duplicate packets are forwarded, our second contribution is an algorithm for video-aware optimized relay selection. Video-awareness is embedded in the relay selection process through a new utility metric that combines the importance of the video packet and the achievable data rate of a particular relay.

Our techniques are fully distributed and require no topology information and exchange of special messages between the relays. Only passively collected local measurements of the channel state are used at the relays.

Section snippets

Related work

Network cooperation for wireless video distribution has been investigated thoroughly. Digital network coding (NC) is a modern technique for packet-level cooperation and it was also studied as a scheme for improving the quality of wireless transmitted video. State-of-art wireless cooperative techniques at the network layer combine algebraic network coding and video transmission [5], [6], [7]. In these works the authors employ linear NC for mixing video packets before transmission to a multicast

System model and assumptions

Small-cell wireless relay network model: In this paper we consider the downlink unicast streaming of a set N{1,2,,n,,N} of N pre-compressed and packetized video flows from a single source node (the MBS), to N destination nodes that each one is interested in a specific video flow. Besides the MBS and the destinations, the network also includes M SCBSs that are not the consumers of the video flows but their task is to aid by forwarding traffic to the destinations as seen in Fig. 1. The main

SCBS overhearing protocol

In this paper our goal is to fully exploit the broadcast property of the wireless channel around the SCBSs which means that we must practically exploit the fact that many SCBSs may receive the same packet. But how do we ensure that duplicate packets are not forwarded from the relays? The first significant component of the proposed system is the overhearing protocol that is responsible for this task. Fig. 3 depicts an example where the contents of the receive buffers at each SCBS are shown.

Distributed video-aware relay selection

Now the relays have a number of available overheard packets, and are also in a position with VPO to know which packets have been transmitted and what was the outcome. So the VPO protocol essentially limits the options for a relay to only the packets that will be useful for a destination. But from all the candidate packets how to select which one to transmit? In the case of digital wireless transmission over slow fading channels when the channel is in deep fade the transmitter must employ

Performance evaluation

In this section, we present a comprehensive evaluation of the proposed algorithms comprising our framework through simulations. We have implemented both the PHY outlined in Section 3, the video streaming system, and the overhearing and relay selection algorithms in Matlab. The number of relay nodes M is kept small since the simulator operates at the PHY symbol level (not packet-level) requiring thus significant amount of execution time. Regarding the lower layer parameters we assume a channel

Conclusions

In this paper we presented a framework for video streaming in dense small-cell wireless networks. Our first contribution was the design of a packet overhearing protocol that exploits the natural diversity that this emerging network paradigm offers. For this system we demonstrated that it is enough to allow overhearing of video packets and employ our distortion optimized streaming approach instead of employing more sophisticated network coding techniques. Interestingly, our overhearing protocol

References (26)

  • J. Andrews

    Seven ways that hetnets are a cellular paradigm shift

    IEEE Commun. Mag.

    (2013)
  • Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, February...
  • Qualcomm. Small cells. [Online]. Available:...
  • Y. Yang et al.

    Relay technologies for wimax and lte-advanced mobile systems

    IEEE Commun. Mag.

    (2009)
  • H. Seferoglu et al.

    Video-aware opportunistic network coding over wireless networks

    IEEE J. Sel. Areas Commun.

    (2009)
  • L. Keller, A. Le, B. Cici, H. Seferoglu, C. Fragouli, A. Markopoulou, Microcast: cooperative video streaming on...
  • S. Chachulski, M. Jennings, S. Katti, D. Katabi, Trading structure for randomness in opportunistic wireless routing,...
  • N. Golrezaei, K. Shanmugam, A. Dimakis, A. Molisch, G. Caire, Femtocaching: wireless video content delivery through...
  • O. Alay, Z. Guo, Y. Wang, E. Erkip, S. Panwar, Enhanced parity packet transmission for video multicast using R-DSTC,...
  • A. Bletsas et al.

    A simple cooperative diversity method based on network path selection

    IEEE J. Sel. Areas Commun.

    (2006)
  • N. Mastronarde et al.

    Transmitting important bits and sailing high radio wavesa decentralized cross-layer approach to cooperative video transmission

    IEEE J. Sel. Areas Commun.

    (2012)
  • K. Poularakis, G. Iosifidis, A. Argyriou, L. Tassiulas, Video delivery over heterogeneous cellular networks: optimizing...
  • Small Cell Forum, Backhaul Technologies for Small Cells: Use Cases, Requirements and Solutions,...
  • Cited by (5)

    View full text