This paper presents an introductory discussion of the factors related to specification and design of bandlimited digital transmission systems. Two classes of band limited signaling pulses are considered: raised cosine and partial response. These signaling pulses are generated by using pulse shaping filters on the output of conventional QPSK, MSK, or QASK modulators. Realization of bandwidth efficient digital transmission systems is reduced to a problem of designing a special class of filters. Performance evaluation indicates information densities of up to 4.2 bits/s/ Hz can be achieved with theoretically no impairment in error performance as compared to the ideal nonbandlimited channel. Computer-aided design methods for realization of the necessary pulse shaping filters are described.

The increasing requirement for high data rate, bandwidth efficient digital radio systems has led to the development of MSK-type modulation methods designed to achieve a compact signal spectrum. These modulation methods include sinusoidal frequency shift keying (SFSK), special MSK-type pulse shapes, and multiamplitude minimum shift keying (MAMSK). When more compact signal spectra are required, bandlimited filtering must be introduced. This note considers the use of conventional MSK or PSK modulators followed by newly developed bandlimited pulse shaping filters. With this approach, it is shown that MSK and offset QPSK modulators yield identical signals on the channel when filtered by properly designed bandlimited pulse shaping networks.

This paper presents a complete statistical characterization of the equilibrium storage requirements for a multiserver queue handling messages with exponential length and service time requirements. The distinguishing feature of the model presented here is that messages are assumed to be stored in unpacked form in finite-size buffers. The model approximates situations frequently encountered in communication network design. The optimum data field size per buffer to result in minimum total storage requirements when considering variable message lengths and buffer overhead is noted and an expression for the buffer utilization is derived.