Multicasting for Efficient Content Delivery

Owning to the popularity of smart mobile devices, the mobile data traffic is growing rapidly. Meanwhile, the type of wireless services is also experiencing a major change, from the traditional voice, e-mail and web browsing, to the emerging video streaming, push media, application download/updates, and mobile TV etc. A central feature of these emerging services is that a same copy of content needs to be delivered to multiple mobile users, referred to as content diversity. An enabling technology to exploit such content diversity is multicasting, as shown in Fig. 1. Compared with point-to-point unicast transmission, point-to-multipoint multicast transmission provides an efficient capacity-offloading approach for common content delivery to multiple subscribers simultaneously on a same resource block. Multicast has also been incorporated in 3GPP LTE standards, known as evolved multimedia broadcast multicast service (eMBMS).

Figure 1: Unicast vs. multicast

Existing multicast transmission is simply to radiate the signals isotropically so that everyone within a coverage area can decode the message. Our research is dedicated to advanced multicast transmission for boosting throughput by exploiting user channel state information as well as content popularity for next-generation mobile communication networks. The specific techniques we are investigating include:

• Coordinated multicast beamforming: We consider the design of multicast beamforming (BF) in future cellular networks as shown in Fig. 2, where each base station applies a single but multi-lobe beamformer to serve concurrently a group of users. Unlike the unicast BF problem in multi-user or multi-cell MIMO systems, which is maturing, the multicast BF problem is generally NP-hard and hence more challenging. We contributed a set of new design methods of coordinated multicast BF in C-RAN architecture. Each base station is capable of inter-cell interference coordination through exchanging user channel state information. Simulation results demonstrated significant advantages of the proposed coordinated multicast beamforming in boosting the worst user-performance and reducing the base station power consumption.

• Massive MIMO multicasting: Massive MIMO is able to significantly improve the spectral/energy efficiency by employing a very large number of antennas. It is considered to be a disruptive technology for 5G wireless communication systems. Our goal is to investigate the benefits of massive MIMO in the context of multicast transmission. We first showed that when each base station (BS) knows the perfect CSI of its own served multicast users, the asymptotically optimal multicast beamformer at each BS has a closed form, expressed as a linear combination of the channel vectors of its multicast users. Then we considered the practical scenario where CSI is obtained through uplink channel estimation in a TDD system. We propose a new uplink pilot scheme that estimates the composite channel rather than each individual user channel. This pilot scheme is able to completely eliminate the so-called pilot contamination problem. Simulation results verified that the proposed scheme works well even at finite number of antennas at each BS.

• Popularity-aware caching and multicasting: Besides multicast transmission for a given set of users who have pre-subscribed a common content, we are currently also investigating adaptive transmission with respect to content popularity by combining caching and multicasting. A basic setup is to deliver M number of contents with different popularity to K users through N base stations, where each content may be cached in different base stations and interested by a different subset of users. The problem is where to catch the contents and how to deliver the contents efficiently.

Figure 2: Coordinated Multicast Beamforming