Many in the communications sector see networked multimedia as the killer app that will jump-start the market. But consumers will need more convincing before they agree to throw away the wireless phones, PDAs or Internet access devices they're using now. And they will be demanding about a raft of issues, including transmission quality and the quality of the voice, data and video output; ease of use; and the network infrastructure.
In many ways, wireless networked multimedia is a microcosm of the larger Net-centric computing and communications environment in which developers must work. Here, they face a bewildering array of market definitions of "multimedia," and questions about how to achieve a desired end in a networked environment with a confusing mix of protocols, mechanisms and standards.
For chip and system designers, the problem is, first, how to deal with the high-performance needs of some types of multimedia processing while remaining sufficiently general purpose to be used in other types of processing. Then, they've got to do it all in as small a footprint as possible and with minimal power consumption and dissipation.
Designers who contributed to this week's In Focus section are looking into a range of alternatives. In his exclusive online contribution, Noel Hurley, consumer entertainment-segment manager at ARM Ltd. (Maidenhead, England), notes that the underlying processor architecture that will satisfy the many requirements of wireless multimedia requires a special mix of DSP, RISC, multiprocessing and SIMD characteristics. And because that mix may vary with the application, Hurley said, ARM's approach is a flexible system-on-chip core that licensees may modify, adding capabilities as functional coprocessing elements. "No single solution is going to optimize a wireless network for highest performance, low cost and power efficiency," he said. "But by combining these high-performance components and methodologies effectively, the cumulative impact on performance is significant."
However, even the most flexible architectures do not deal with the realities of design in this still volatile environment, said contributor Larry Przybylski, principal software engineer at Media Works Technology Corp. (Newport Beach, Calif.). It often takes many iterations to determine the right mix for a particular wireless-multimedia segment.
"Sending still images over Bluetooth to and from a network-enabled digital camera; moving video and synchronized voice for a high-end 3G/4G cell phone; performing mixed graphics, animation and voice in an existing cell phone-all require different mixes of instructions and hardware capabilities," Przybylski said. For that reason, Media Works uses configurable processors from companies such as Altera and Tensilica.
But the complexity does not stop at the device, board or system I/O. Much depends on the software and protocol infrastructure. In that respect, wireless multimedia still has a way to go before it becomes a reliable delivery mechanism, said Vijay Madisetti, CEO of SoftNetworks LLC (Atlanta). Aside from limited bandwidth, the medium itself is subject to problems such as multipath fading, noise susceptibility, and high packet loss and bit error rates.Eye on transport
The heterogeneous 3G infrastructure and the inadequate performance of transport-layer protocols also need to be resolved. Designed for data networks, TCP and UDP are "not expected to perform satisfactorily in the wireless content delivery network scenario," he said. There is also a need, said Madisetti, for authentication and encryption schemes that combine performance with low power requirements.
In his exclusive online article, Gene Eagle, senior software architect at Intel Corp., explains the importance of enhancements to the Extensible Markup Language-VoiceXML and Speech Application Layer Tags-in the initial phases of wireless multimedia.
Efforts are under way to solve a number of interoperability issues that plague wireless multimedia. For instance, an MPEG-2 video bit stream may have to be converted to MPEG-4 over wired and then wireless media; or an MPEG-2 video originally encoded at 5 Mbits/second might have to be converted to 3 Mbits/s or lower for wireless links.
There is also the problem of conversion to and from heterogeneous Internet environment and the similarly diverse wireless one. "While we all dream that voice (and video) standards will converge to a single standard across wireless and wireline networks, the reality is that legacy networks must continue to be supported," said contributor Marwan Jabri, CTO of Dilithium Networks Inc. (Larkspur, Calif.). Currently, mechanisms for translating or transcoding across the various networks introduce delays that prevent successful, high-quality wireless multimedia communications, he said.