How to Deliver Layer-Coded Content over Broadcast Channels: An Information-Theoretic Perspective


Broadcast multicast service (BCMCS) has been increasingly popular for delivering multimedia contents to mobile users. BCMCS can be implemented either through a dedicated digital broadcast infrastructure such as DVB-T/H/S2, MediaFLO and DMB or through a mobile network such as UMTS and cdma2000 network. Traditional digital broadcast air interfaces and networks are designed with consideration on the tradeoff between maximum achievable rate and the intended coverage. The actual rates are usually limited by the maximum transmit power and the worst channel condition so that every user in coverage can reliably receive the same quality and content of services. It means that the users under good reception condition may have no advantage, even if their potential throughputs can be much higher. This situation frequently happens especially to mobile users, whose reception conditions change all the time. Therefore, digital broadcast service can become more attractive if we can provide users with differentiated qualities of services (QoS’s) based on the contents they requested, the receiver capability, and reception conditions. On the other hand, there are also arising interests in upgrading existing digital broadcast systems by introducing more services and delivering more QoS options to users with advanced receivers while still guaranteeing existing users' services and providing unequal error protection on content deliveries. Many technologies are under investigation for these goals, e.g., rateless coding, hierarchical modulation, multiple-input multiple-output, selective retransmission and superposition precoding (SPC). Since there are a large number of users already served by existing systems, it is prohibitively expensive to simply replace their existing user devices by next-generation ones. Backward compatibility and implementation complexity are among the major concerns in upgrading existing systems. It is expected that existing receivers are able to continue to operate in upgraded systems, even though they may not be able to receive the newly provided services. In addition, there always are power consumption and computation complexity concerns in mobile device design.

Among the technology candidates, SPC is one of the most promising technologies for upgrading existing systems with larger coverage and more QoS options while maintaining strictly backward compatibility. It was developed to achieve the maximum sum rate of Gaussian broadcast channels with superpositioning two independent signals at the transmit side and employing interference cancellation at the receive side. One of the key advantages of SPC is that the added complexity and cost are low. The most popular SPC implementation is hierarchical modulation, also called layered modulation, where multiple data streams are multiplexed and modulated into one single symbol consisting of base-layer subsymbols and enhancement-layer subsymbols. It has been widely proven and included in various standards such as DVB-T, MediaFLO, UMB (Ultra Mobile Broadband, a new 3.5th generation mobile network standard developed by 3GPP2), and is under study for DVB-H. However, it is known that hierarchical modulation suffers from the interlayer interference (ILI) between the lower-layer signals and high-layer signal(s). For example, for a hierarchically modulated symbol with 16QAM base layer and QPSK enhancement layer, the base-layer throughput loss can be up to about 1.5 bits/symbol at around 23 dB due to ILI. This is about 1.5/4 = 37.5% of the base-layer achievable throughput. ILI increases the demodulation error rate of higher-layer symbols too. In addition, there is no interlayer diversity in hierarchical modulation since the data stream of each layer experiences the exact same channel condition. Therefore, superimposed OFDM (S-OFDM), an alternative SPC implementation, is propose to provide higher spectral efficiency and additional flexibility with less ILI and additional interlayer diversity gain.

S-OFDM implements SPC by precoding and superimposing two independent OFDM signals together. One typical S-OFDM signal consists of at least two layer OFDM signals, where the base layer can be an existing OFDM signal and the enhancement layer is a precoded OFDM signal, which is usually of the same numerology but with additional symbol spreading across subcarriers using, e.g., Walsh-Hadmard matrix. From an information-theoretical perspective, the achievable throughput is higher because the signal constellation size of each resulted subcarrier becomes larger and the ILI is minimized too. From a signal processing perspective, there is additional interlayer diversity between these two layers because the channel experienced by each layer becomes different to each other due to the precoding. This means, if the base layer signal goes through a bad channel, there still is a chance for the enhancement layer to have a good channel. It is shown that more interlayer diversity can result in more robust transmission and higher achievable spectral efficiency. This is very important for BCMCS liked applications where there usually is no feedback channel. Compared with many other techniques, the advantages of the proposed scheme include the incurred implementation complexity is low and it is fully backward compatible with small change on existing systems.


[3GPP2 NTAH 2008: Precoded OFDM for BCMCS Evolution, Xiamen, Fujian, China]