Bergmans and Cover identified the capacity region of the Gaussian degraded broadcast channel, where different receivers observe the transmitted signal with different signal-to-noise ratios. This letter presents a superposition turbo-coding scheme that performs within 1 dB of the capacity region boundary of the degraded broadcast channel at a bit-error rate of 10/sup -5/.

Bergmans and Cover identified the capacity region of the Gaussian degraded broadcast channel, where different receivers observe the transmitted signal to noise ratios. This paper presents a superposition turbo coding scheme that performs within 1 dB of the capacity region boundary of the degraded broadcast channel at BER of 10/sup -5/. Performance is consistent over the entire useful range of the power allocation parameter /spl alpha/. Coding for the degraded broadcast channel is equivalent to coding for unequal error protection. Adjusting /spl alpha/ changes the transmitter constellation of our encoder, and changes the degree to which error protection is unequal. When /spl alpha/ is selected to provide equal error protection, the code is essentially a multilevel turbo trellis-coded-modulation schemes with the advantage of the potential for flexible unequal error protection as /spl alpha/ is varied.

We present a technique to find the optimal threshold /spl tau/ for the binary hypothesis detection problem with n identical and independent sensors. The sensors all use an identical and single threshold /spl tau/ to make local decisions, and the fusion center makes a global decision based on the n local binary decisions. For generalized Gaussian noise and some non-Gaussian noise distributions, we show that for any admissible fusion rule, the probability of error is a quasi-convex function of threshold /spl tau/. Hence, the problem decomposes into a series of n quasi-convex optimization problems that may be solved using well-known techniques. Assuming equal a priori probability, we give a sufficient condition of the non-Gaussian noise distribution g(x) for the probability of error to be quasi-convex. Furthermore, this technique is extended to Bayes risk and Neyman-Pearson criteria. We also demonstrate that, in practice, it takes fewer than twice as many binary sensors to give the performance of infinite precision sensors in our scenario.