Apparatus for Monitoring an ATM System

The invention relates to an apparatus for determining a measured value during the arrival of an ATM cell which forms part of an ATM cell flow in an ATM system. Such an apparatus is disclosed, inter alia, in European Patent Application EP-381 275 in the name of the Applicant. This is used to protect an ATM system against impermissible transgressions of the nominal (maximum) cell rates assigned to the cell flows (virtual channels) provided. The known apparatus makes use of a so-called `leaky bucket` mechanism. A problem which occurs in the transmission of ATM cells is that, as a consequence of the individual manipulability thereof, the cells may have a fairly large variation in arrival time (cell delay variation). Insofar as this variation is a consequence of manipulations in the transmission system itself (such as in switches and buffers) (see, inter alia, European Patent Application EP-450 708 in the name of the Applicant), this must not result in the cells being rejected. On the other hand, cell series which do (substantially) exceed the cell rate through the fault of the users are in fact refused further transmission (or, for example, marked as in European Patent 0 275 678 in the name of AT&T). The said `leaky bucket` mechanism has too few degrees of freedom to monitor the cell flow accurately enough for it to be possible to deduce with a reasonable probability from the measured value obtained whether a (temporary) cell rate increase is the consequence of an unavoidable variation induced by the transmission system or the consequence of an unduly high cell rate (which is in fact avoidable) on the part of the system users.

The object of the invention is therefore to provide an improved apparatus for determining a measured value during the arrival of an ATM cell which forms part of an ATM cell flow in an ATM system, which apparatus is characterized in its generality in that the measured value (CV.sub.n) is a function of a reference arrival time (RAT.sub.n) assigned to said arriving cell and the actual arrival time (AT.sub.n) of said cell, which reference arrival time (RAT.sub.n) is a function of the reference arrival rime (RAT.sub.n-p) of at least one preceding cell and of the actual arrival time (AT.sub.n-p) of said same at least one preceding cell, these features expressed as a formula being:

In this way, only the reference arrival time (RAT.sub.n-p) and the actual arrival time (AT.sub.n-p) of a preceding cell (or, possibly, a plurality of preceding cells, for example, the cells n-1 to n-5) need to be temporarily stored in a buffer in all cases, while all the actual arrival times of the entire cell flow nevertheless exerts an influence on the reference arrival time of the last cell to arrive and, consequently, on the required measured value, since each reference arrival time is calculated from a preceding actual arrival time and a preceding reference arrival time; said preceding reference arrival time is calculated from an actual arrival time and a reference arrival time situated in front thereof, etc. The reference arrival time serves, as it were, as transfer medium for propagating the influence of the past (actual) arrival times up to the present. In general, as is stated above, the parameters which regulate the influence of the past arrival times on the most recent reference arrival time will be chosen so that the arrival times of old cells have less influence on the most recent reference arrival time than the younger cells.

In fact, it is not important whether the said arrival times are expressed in absolute values, for example related to one and the same starting value (T=0) or whether they are expressed in mutually related interval times.

If the arrival times are expressed in absolute values, a more detailed elaboration of the invention is preferably characterized in that the measured value (CV.sub.n) is proportional to the difference between the reference arrival time (RAT.sub.n) and the actual arrival time (AT.sub.n) of the arriving cell, and the reference arrival time (RAT.sub.n) is proportional to the reference arrival time (RAT.sub.n-p) of a preceding cell and proportional to the actual arrival time (AT.sub.n-p) of that same preceding cell and is multiplied by the inverse value (T) of a nominal cell rate assigned to the ATM cell flow, these features expressed as a formula being:

CV.sub.n =C.sub.1 *(RAT.sub.n -AT.sub.n), where RAT.sub.n =C.sub.2 *RAT.sub.n-p +C.sub.3 *AT.sub.n-p +T, where C.sub.1, C.sub.2 and C.sub.3 are proportionality factors.

If the arrival times are expressed in mutually related values, a more detailed elaboration of the invention is preferably characterized in that the measured value (CV.sub.n) is proportional to the difference between the reference arrival time (RAT.sub.n) and the actual arrival time (AT.sub.n) of the arriving cell, and the reference arrival time (RAT.sub.n) is proportional to the measured value (CV.sub.n-p) of a preceding cell and is multiplied by the inverse value (T) of a nominal cell rate assigned to the ATM cell flow, these features expressed as a formula being:

CV.sub.n =C.sub.1 *(RAT.sub.n -AT.sub.n), where RAT.sub.n =C.sub.2 *CV.sub.n-p +T, where C.sub.1 and C.sub.2 are proportionality factors.

In both cases, the (absolute or relative) reference arrival time of a cell are determined from the reference arrival time and the actual arrival time of a preceding cell (for example the last cell (p=1)). Since the reference time of the preceding cell was again also determined from a preceding cell (etc., etc.) it is now possible to arrange for the `arrival-time history` of the cell flow to play a pan in assessing the behaviour of the cell flow and conclusions can be drawn therefrom regarding the cause of (shorter or longer) transgressions of the nominal cell rate. The degree to which older and younger preceding cells exert an influence on the most recent reference arrival time is determined by the proportionality factors C.sub.2 and C.sub.3. It is also possible to arrange for said factors C.sub.2 and C.sub.3 to vary automatically under the influence of, for example, the average arrival time deviation or other variables derived from the behaviour of the cell flow. For example, in this way, a change in behaviour detected at a certain instant in the cell flow can be investigated more accurately by altering the detection parameters, proportionality factors C.sub.2 and C.sub.3, after detecting said change in behaviour; it is therefore possible to `zoom in`, as it were, on said change in behaviour, as a result of which a more accurate conclusion can be drawn with regard to its cause.

FIG. 1 shows diagrammatically an exemplary embodiment of a measuring and monitoring apparatus according to the invention, in which the method of the invention is implemented. Use is made of the absolute arrival times.

FIG. 2 shows a listing explaining calculating CV based on cell arrival times, where RAT=Reference Arrival Time; AT=Arrival Time; ATD=Arrival Time Deviation; CV=Control Value; and TH=Threshold; and

FIG. 3 shows a listing explaining calculating CV based on cell arrival intervals, where RAI=Reference Arrival Interval; AI=Arrival Interval; AID=Arrival Interval Deviation; CV=Control Value; and TH=Threshold.

The measuring and monitoring apparatus shown in FIG. 1 comprises a read-out device 1, a switching device 2, six memory devices 3, 5, 7, 9, 11 and 13, a time-indicating device 4 and a processor comprising four calculating devices 6, 10, 12, 14 and a comparison device 8.

At an input A of the measuring and monitoring apparatus, there arrive sequentially data cells which comprise a control word group and an information word group. Located at that point is the read-out device 1 which derives the channel number from the control word group. The data cell itself is passed to the switching device 2, where the data cell is temporarily stored until a decision is taken about whether or not to pass said data cell to output B of the switching device 2, to which a downstream transmission network (not shown) is connected.

The channel number derived by 1 serves as an index with which the memory devices 3, 5, 7, 9, 11, 13 deliver the parameters and variables which are specific to the channel concerned. The first memory device 3 provides an indication of the resolution with which the time has to be determined by the time-indicating device 4. The second memory device 5 delivers an indication of the reference time of the data cell with the respective channel number as calculated after expiry of the preceding data cell having that same channel number. The first calculating device 6 then determines what the difference is between the instant of arrival a(n) of the last data cell to arrive and the instant of reference c(n) valid for said cell. In the case of measurements, the measured value thus obtained can be outputted for storage, for example together with the channel designation of the cell.

The reference time is then determined as follows for the subsequent data cell with the respective channel number. A first constant value C.sub.2 is supplied by the fourth memory device 9. In the second calculating device 10, said value is multiplied by the reference time RAT.sub.n previously retrieved from the memory device 5. The fifth memory device 11 supplies a second constant value C.sub.3. In the third calculating device 12, this value is multiplied by the arrival time AT.sub.n previously obtained from the time-indicating device 4. The sixth memory device 13 supplies a third constant factor T. This value is summed in the fourth calculating device 14 with both the result from calculating device 10 and with that from 12. The result of 14 is the reference time RAT.sub.n+1 for the subsequent data cell having the same channel number.

The comparison device 8 compares the measured value obtained from calculating device 6 with a threshold value TH supplied by the third memory device 7.

If the comparison device concludes that the result is less than (or equal to) the threshold value, the following actions are taken:

(a) The switching device 2 is instructed to pass the respective data cell (via its output B) to the downstream transmission network.

(b) The value obtained from calculating device 14 for the reference time for the subsequent data cell having the same channel number is written into the second memory device 5, the old value in said memory device being overwritten.

If the comparison device concludes that the result is greater than the threshold value, the following actions are carried out:

(a) The switching device 2 is instructed not to pass the respective data cell to the downstream transmission network. In many implementations of networks, it is required in such a case that the vacated position is replaced by a certain bit pattern which indicates, for example, a blank data cell; the switching device 2 then transmits (via its output B) said bit pattern instead of the respective data cell.

(b) The value obtained from calculating device 14 for the reference time is not written into the second memory device 5 so that the old value is retained in said memory device.

During initialization of the apparatus, or of a transmission channel, the memory devices 3, 5, 7, 9, 11, 13 are filled from the control system (not shown) with parameters specific to the various channel numbers, along with the initial values (starting values) for the variable in the second memory device 5. For this purpose, the memory devices are provided with an adjustment facility which gives the control system access to the memory means without interfering with the actual operation of the measuring and monitoring apparatus.

If the method is used solely as a measuring instrument and not as a monitoring apparatus, the switching device 2, the third memory device 7, the comparison device 8 and the control system for preventing the storage of the calculated instant of reference in memory device 5 can be omitted.

FIGS. 2 and 3 show the result of a simple simulation of the invention for a series of 50 cells which enter the apparatus sequentially. The programs used to generate these results are shown by Algorithms 1 and 2; the program shown in Algorithm 1 makes use of the absolute values of the arrival times, indicated by AT (arrival time) and RAT (reference arrival time); the program structure corresponds to the structure of the exemplary hardware embodiment in FIG. 1. The program shown in Algorithm 2 makes use of the detected arrival times with respect to one another, indicated by AI (arrival interval) and RAI (reference arrival interval). Although the intermediate results are different, the final results are identical. An exemplary hardware embodiment such as that in FIG. 1 can be assembled without great effort from the structure of the program shown in Algorithm 2, but making use in that case of arrival time intervals.