International Business Machines Corporation

Mehdi M Nassehi

Erwin A Zurfluh

Werner K Bux

Johann R Mueller

Pitro A Zafiropulo

Peter L Heinzmann

Johann R Muller

Douglas H Dykeman

Willibald A Doeringer

Mohan B Sharma

Gunter Karjoth

A computerized method, encoded on a computer-readable medium, of detecting anomalies in an event stream. The method comprises at least two acts. In a first act, the method uses a tree structure to extract a grammar having an associated set of rules, from a sample of normal behavior. In a second act, the method checks an event stream against the rules of the grammar to detect anomalies.

Anomaly detection method

Variable length data sequence backtracking a trie structure

There is provided a method of regulating access to a unidirectional bus transmission system to which a plurality of stations identified by labels are connected, on which system data are transmitted in time slots which are released in numbered cycles by a headend unit, and in which said headend unit sends, for each of said cycles, a request poll message carrying the respective cycle number. The method comprises i) logically combining a source station label and a destination station label, ii) deriving a first control number from the logical combination of the source station label and the destination station label, iii) using, for requesting slots in which to transmit said data, a request poll message carrying a cycle number which corresponds to said first control number, and iv) subsequently transmitting said data in slots of a cycle whose number corresponds to said first control number.

Communications network and method of regulating access to the busses in

A protocol and format for multiple accessing and transporting information in a Distributed Queue Dual Bus communications network includes count field in which stations can reserve multiple time slots for transporting data.

Method and apparatus for distributed queue multiple access in a

In a communications network comprising two counterflowing busses (folded bus or dual bus) and a number of stations (N1, N2, N3) connected between them, a cyclic reservation and access technique is provided. A headend unit generates consecutive time slots for data transmission which are grouped in cycles. For each cycle, the headend first issues a reservation command (RES) with a cycle number and a reservation count which each station augments by the slot count required locally. Each station stores the cycle number with the requested slot count. The headend stores in a global reservation queue, for each returning reservation command, the cycle number and the accumulated reservation count. Later it issues a start command (ST) with the respective cycle number, and subsequently generates the required number of slots as indicated by the stored accumulated reservation count. The headend can issue additional confirm commands and reject commands for activating a backpressure mechanism.

Method and apparatus for cyclic reservation multiple access in a

In a communication system comprising node stations connected to a bus, a headend unit releases, for sequentially numbered cycles, time slots as previously requested in an order pad passing procedure. In the order pad procedure, the headend issues a special frame in which stations insert the number of slots required to transmit data in a subsequent cycle. To enable reuse of time slots which otherwise would be utilized only during a fraction of their existence (i.e. from a source to a destination which in many cases are located close together), the order pad additionally includes a furthest destination label (DST-MAX) and a requested maximum (REQ-MAX) which are updated at each node. This allows to set, for any cycle, at some nodes a reuse flag (REU-FLG). In this way, the bus is partitioned into sections according to actual traffic, and the headend only releases a number of time slots which satisfies the needs of the section where the maximum of time slots was requested. Thus, the number of time slots, i.e. the length of each cycle can be significantly reduced, thereby increasing system throughput.

Multiple-access control for a communication system with order pad passing

A distributed queue multiple access technique is disclosed for a system comprising two counterflowing transmission busses (A,B), plural stations (i, j) connected between them, and a headend station generating slots on the first one (A) of the busses; each station transmitting access requests on the second one (B) of the busses and determining its right to transmit data in slots on the first bus on the basis of access requests it has seen. Each station maintains a FIFO queue (41) for sequential storage of numbers representing external (EXT-REQ) and local (LOC-REQ) access requests. To achieve correct timing between access requests entering the queue from the second bus and respective free slots passing a station on the first bus, a station-individual delay is introduced by maintaining in a FIFO delay storage (71) a specific number of entries. This results in an equally fair treatment for all stations, and guarantees the availability of consecutive free slots for any station which requested them.

Growing community of inventors

Adliswil, Switzerland

Mehdi M Nassehi