CHANGES IN PREDICTED MOS DEVICE CHARACTERISTICS DUE TO THE USE OF TWO DIMENSIONAL SOURCE/DRAIN PROFILES.

Two-dimensional MOS device simulation programs such as MINIMOS left bracket 1 right bracket are limited in their validity due to assumptions made in defining the initial two-dimensional source/drain profiles. The two options available to define source/drain regions both construct a two-dimensional profile from one-dimensional profiles normal to the surface. Inaccuracies in forming these source/drain profiles can be expected to effect predicted device characteristics as channel dimensions of the device are reduced. This paper examines these changes by interfacing numerically similated two dimensional source/drain profiles to MINIMOS and comparing predicted I//D-V//D characteristics with 2-D interfacing, 2-D profiles constructed from interfaced 1-D profiles and MINIMOS self generated profiles. Data obtained for simulations of 3 mu m N and P channel devices are presented.

Gaussian Process training with input noise

In standard Gaussian Process regression input locations are assumed to be noise free. We present a simple yet effective GP model for training on input points corrupted by i.i.d. Gaussian noise. To make computations tractable we use a local linear expansion about each input point. This allows the input noise to be recast as output noise proportional to the squared gradient of the GP posterior mean. The input noise variances are inferred from the data as extra hyperparameters. They are trained alongside other hyperparameters by the usual method of maximisation of the marginal likelihood. Training uses an iterative scheme, which alternates between optimising the hyperparameters and calculating the posterior gradient. Analytic predictive moments can then be found for Gaussian distributed test points. We compare our model to others over a range of different regression problems and show that it improves over current methods.

Second-order rate region of constant-composition codes for the multiple-access channel

This paper presents an achievable second-order rate region for the discrete memoryless multiple-access channel. The result is obtained using a random-coding ensemble in which each user's codebook contains codewords of a fixed composition. It is shown that this ensemble performs at least as well as i.i.d. random coding in terms of second-order asymptotics, and an example is given where a strict improvement is observed. © 2013 IEEE.