On Distributed Communications Series
IV. Priority, Precedence, and Overload
In standard military communications systems, there is a natural upgrading of message precedence in times of stress. But, during the 99 per cent of the time of little or no stress conditions, the circuits are not conditioned for this high-stress traffic.
The Distributed Adaptive Message Block Network system delivers all traffic within the same time constraints, and that amount of time is within the limits of present emergency-level requirements. The circuits are prevented from overloading only by restriction of input--and no complete cutoff of input occurs to any user.
This Memorandum considers four separate techniques that can be used singly or in combination to achieve, through automation, "best" use of a seriously degraded and overloaded communications plant, within the framework of a rapidly changing organizational structure.
In the schemes considered, precedence is determined moment-by-moment, automatically for all traffic in the network. Precedence is computed as a composite function of:
- the ability of the network to accept additional traffic;
- the "importance" of each user and the "utility" of his traffic;
- the data rate of each input transmission medium or the transducer used;
- the tolerable delay time for delivery of the traffic.
During overload conditions, precedence status information is fed back through the network to limit the type and volume of data flow allocated to each user.
The dual goal is to prevent complete denial of communications to any network user while preventing network overloading. The surviving data rate, however meager under heavy network degradation, is always "equitably" rationed among the many network users. The definitions of "equitable" and "important" change from time to time, with control reserved to the commanding authority.
The examples used for describing the proposed notions are based upon the use of a high-data-rate, time-division, distributed system whose flexibility permits a rapid exchange of channel allocation between many users using few bits per second, and a few users using many bits per second. While the automated computational apparatus needed could, in all probability, be practicably implemented only in a future communications network, the concept is quite general and is easily visualized as being applicable to a pile of papers on a busy executive's desk, to a stack of computer programs awaiting processing, or to any other form of communications system.