Reliable file distribution over mobile broadcast systems

摘要

The 3GPP Multimedia Broadcast and Multicast Service (MBMS) provides new bearer services and procedures for efficient transmissions to large user groups. When the group is large, MBMS distributes content by using broadcast on the air-interface. This thesis evaluates the use of MBMS for reliable file distribution services. One important requirement for file distribution is that the files contain no transmission errors. The MBMS file distribution process is subdivided into two phases in this thesis: During the first phase, the radio access network sends the IP packets in each cell either using one broadcast channel or several ptp channels depending on the number of receivers. During the second phase, the file repair service is executed when needed. The file repair service uses either HSPA bearers or MBMS bearers. It is mandatory when minimizing the needed resources for reliable file transmission. In order to understand the transmission characteristics of the first phase, we analyze the packet transmission over the MBMS traffic channel (MTCH). The use of shorter IP packets leads to a lower IP packet error probability on the MTCH. When using shorter IP packets, a larger share of bits is spent on packet headers. To evaluate the information throughput over MTCH, we define the goodput as the fraction between received information bits and sent data bits. IP packets smaller than 500 Byte lead in case of block error rates larger than 10% to a higher goodput. We evaluate different optimization targets for MBMS file delivery. The most important evaluation target is to balance both transmission phases. The resource usage for the MBMS transmission is balanced with the resource needs for the file repair in order to increase the system efficiency of the file distribution of a certain size to all receivers. It is possible to trade the transmit power with the amount of application layer FEC redundancy at same load for the file repair service. The Raptor FEC is used for MBMS. Additional FEC redundancy increases the needed transmission energy, since the system resources are used for a longer time. The point-to-point file repair uses unicast HTTP connections and spreads the repair requests in a time window. The receivers draw randomly a start time out of a given wait-time window. The link between the file repair server and the system limits significantly the serving time and may even lead to an under utilization of the radio resources. The PTP file repair is well dimensioned when the radio links of all active file repair receivers and the link to the file repair server are just fully utilized. The smallest file repair service duration takes approximately 1.2 times the Sequential Delivery Time of all missing data over the link between the file repair server and the system.