The method of passive beam formation using a four-element Butler matrix to improve the signal availability of meteor scatter communication systems is investigated. Signal availability, defined as the integrated time that the signal-to-noise ratio (snr) exceeds some sf•r threshold, serves as an important indicator of system performance. Butler matrix signal availability is compared with the performance of a single four-element Yagi reference system using --6.5 hours of data from a 720 km north-south temperate latitude link. The signal availability improvement factor of the Butler matrix is found to range between 1.6-1.8 over the snr threshold range of 20-30 dB in a 300-Hz bandwidth. Experimental values of the Butler matrix signal availability improvement factor are compared with analytical predictions. The experimental values show an expected snr threshold dependency with a dramatic increase at high snr. A theoretical analysis is developed to describe this increase. The signal availability can be further improved by --10-20% in a system employing two four-element Butler matrices with squinted beams so as to illuminate the sky with eight high-gain beams. Space diversity is found to increase the signal availability of a single antenna system by 15%, but the technique has very little advantage in a system already employing passive beam formation.One strategy to overcome the low data throughput focuses on improved error coding. Fixed rate forward error correction (FEC) coding offers an improvement of--2•% in the data throughput over uncoded systems [Miller and Milsrein, 1990]. Further slight improvements can be made by using more complex variable-rate coding schemes [Pursley and Sandber•, 1989], but significant advances are constrained by the limited bandwidth of most M$C systems. An alternative technique is •o employ variable data transmission rates, whereby the data rate profile mimics, in some fashion, the decay profile of the trail snr. Such a scheme takes advantage of the Published in 1996 by the American Geophysical Union. Paper number 95RS03258. relatively high snr at the beginning of the trail [e.g., Weitzen et al., 1984; Davidovici and Kanterakis, 1993]. Shukla et al. [1992] and Cannon et al.[1993] investigated increasing the trail duration, and thereby the data throughput, by using antenna space diversity. This technique takes advantage of the fact that toward the end of a trail, and particularly in the case of long-duration trails, the incoming wavefront is no longer planar because of interference between decorrelated signals. Shukla et al. [1992] found that -40% of all trails with duration >0.75 s showed significant signs of spatial decorrelation over antenna separations of 10A. Further breakdown of trail statistics into underdense, overdense, and not-known categories showed that the underdense and not-known categories did not benefit from increasing the antenna separation above 5A. The overdense trails, however, required a separation of 20A before signal decorrelation was apparent. Cannon et al. [1993] ...