Mobile Supercomputers for the Next-Generation Cell Phone
FREE IEEE PAPERS
Mobile devices have pr olif er a t e d a t a spectacular rate, with more than 3.3 billion active cell phones in the world. Soon, improvements to today’s smart phones, such as high-bandwidth Internet access, high-definition video processing, and interactive video conferencing will be commonplace. The Int er na t iona l Tele communications Union has proposed fourth-generation (4G) wireless technology to increase bandwidth to maximum data rates of 100 Mbps for high-mobility situations and 1 Gbps for stationary and low-mobility scenarios like Internet hot spots (www. research802.org/secmail/pdf00204.pdf). This translates into an increase in computational requirements of 10 to 1,000 times over previous thirdgeneration (3G) wireless technologies, with a power budget of approximately 1 W for all the computation. Other forms of signal processing, such as high-definition video, are also up to 100 times more compute-intensive than current mobile video. Figure 1 shows the peak processing throughputs and power budgets of 3G and 4G protocols. Conventional processors cannot meet these protocols’ powerthroughput requirements. Research solutions, such as VIRAM and Imagine, can achieve the performance requirements for 3G, but generally exceed the power budgets of mobile terminals. The signalprocessing on demand architecture (SODA) improved upon these solutions and could meet both the power and throughput requirements for 3G wireless
For 4G wireless protocols, the computational efficiency of mobile computer systems must be increased to greater than 1,000 Mops/mW. 4G uses three central technologies: orthogonal frequency-division multiplexing (OFDM), low-density parity check (LDPC) code, and multiple-input multiple-output (MIMO) techniques.