The document discusses different approaches to implementing stored program control in electronic switching systems, including centralized and distributed control. Centralized control uses a single processor to control the entire exchange, while distributed control uses multiple processors that work independently or share load. Fault tolerance is an important requirement for reliable telephone switching.
The document discusses different approaches to implementing stored program control in electronic switching systems, including centralized and distributed control. Centralized control uses a single processor to control the entire exchange, while distributed control uses multiple processors that work independently or share load. Fault tolerance is an important requirement for reliable telephone switching.
Early crossbar systems were slow in call processing as
they used electromechanical components for common control subsystems. Efforts to improve the speed of control & signalling between exchanges led to the application of electronics in the design of control & signalling subsystems. In late 1940s & early 1950s, vacuum tubes, transistors, gas diodes, magnetic drums & CRTs are used for realizing control functions. Contemporary to these developments was the arrival of modern electronic digital computers. Switching engineers then replaced the registers and translators of the common control systems by a single digital computer 1 Stored Program Control In stored program concept, a program or a set of
instructions to the computer is stored in its memory
and the instructions are executed automatically one by one by the processor. Carrying out the exchange control functions through
programs stored in the memory of a computer led to
the nomenclature stored program control (SPC). An immediate consequence of program control is
the full-scale automation of exchange functions &
the introduction of a variety of new services to users like common channel signalling (CCS), centralized maintenance & automatic fault diagnosis, interactive human-machine interface etc. 2 Stored Program Control A telephone exchange must operate without interruption, 24 hours a day, 365 days a year & for say, 30-40 years. This means that the computer controlling the exchange must be highly tolerant to faults. Early commercial computers were not fault tolerant. In fact, major contributions to fault tolerant computing have come from the field of telecommunication switching. The world’s first electronic switching system, known as No. 1 ESS, was commissioned by AT&T in May 1965. However, attempts to replace the space division electromechanical switching matrices by semiconductor crosspoint matrices have not been greatly successful. As a result, many space division electronic switching systems use electromechanical switching networks with SPC. 3 Stored Program Control Both the types qualify as electronic switching systems although only one of them is fully electronic. With the evolution of time division switching, which is done in the electronic domain, modern exchanges are fully electronic. There are basically two approaches to organizing stored program control: Centralized Distributed 4 Centralized SPC In centralized control, all the control equipment is replaced by a single processor which must be capable of processing 10- 100 calls per second & simultaneously performing many other ancillary tasks. A centralized SPC configuration may use more than one processor for redundancy purposes. Each processor has access to all the exchange resources like scanners and distribution points & is capable of executing all the control functions. 5 Centralized SPC In actual implementation, the exchange resources and the memory modules may be shared by processors, or each processor may have its own dedicated access paths to exchange resources & its own copy of programs and data in dedicated memory modules. Present day ESS using centralized control uses only a two-processor configuration. 6 Centralized SPC A dual processor architecture may be configured to operate in one of three modes: Standby mode Synchronous duplex mode Load sharing mode In standby mode, normally, one processor is active & the other is on standby, both hardware & software wise. The standby processor is brought online only when the active processor fails. An important requirement of this configuration is the ability of the standby processor to reconstitute the state of the exchange system when it takes over the control. In small exchanges, this may be possible by scanning all the status signals as soon as the standby processor is brought into operation. In such a case, only the calls which are being established at the time of failure of the active processor are disturbed. 7 Centralized SPC In large exchanges, the active processor copies the status of the system periodically into a secondary storage. When a switch-over occurs, the online processor loads the most recent update of the system status from the SS & continues the operation. Only the calls which changed status between the last update and the failure of the active processor are disturbed. 8 Synchronous duplex mode In SDM, hardware coupling is provided between the two processors which execute the same set of instructions and compare the results continuously. If a mismatch occurs, the faulty processor is identified and taken out of service within a few milliseconds. When the system is operating normally, the two processors have the same data in their memories at all times & simultaneously receive all information from the exchange environment. One of the processors actually controls the exchange and the other is synchronized with the former but does not participate in the exchange control. If a fault is detected by the comparator, the two processors are decoupled and a check-out program is run independently on each of the machines to determine which one is faulty. When a faulty processor is repaired & brought into service, the memory contents of the active processor are copied into its memory & brought into synchronous operation with the active processor & then the comparator is enabled. 9 Synchronous duplex mode It is possible that a comparator fault occurs on account of a transient failure which does not show up when the check-out program is run. In such cases, the decision as to how to continue the operation is arbitrary and three possibilities exist: Continue with both the
processors Take out the active
processor & continue with
the other processor Continue with the active
processor but remove the
other processor 10 Load sharing operation In load sharing operation, an incoming call is assigned randomly or in a predetermined order to one of the processors which then handles the call right through completion. Thus, both the processors are active simultaneously and share the load & the resources dynamically. Since the calls are handled independently by the processors, they have separate memories for storing temporary call data. Programs & semi- permanent data are also kept in separate memories for redundancy purposes 11 Load sharing operation There is an interprocessor link through which the processors exchange information needed for mutual coordination and verifying the state of health of the other. If the exchange of information fails, one of the processors which detects the same takes over the entire load including the calls that are already set up by the failing processor. Sharing of resources calls for an exclusion mechanism so that both the processors do not seek the same resource at the same time. In hardware implementation, the exclusion device which, when set by one of the processors, prohibits access to a particular resource by the other processor until it is reset by the first processor. Load sharing configuration increases the effective traffic capacity by about 30% when compared to SD operation. LS is a step towards distributed control. 12 Load sharing operation One of the main purposes of redundant configuration is to increase the overall availability of the system. A telephone exchange must show more or less a continuous availability over a period of perhaps 30 to 40 years. Let’s compare the availability figures of a single processor & a dual processor system. The availability of a single processor system is given by, A=MTBF/(MTBF+MTTR) where, MTBF= mean time between failure MTTR= mean time to repair 13 Load sharing operation The unavailability of the system is given by, U= 1-A=MTTR/(MTBF+MTTR) If MTBF>> MTTR, Then, U=MTTR/MTBF For a dual processor system, MTBFD, can be computed from the MTBF & MTTR values of the individual processors. A dual processor system is said to have failed only when one of the processors has failed and the second processor also fails while the first one is being repaired. In other words, this is related to the conditional probability that the second processor fails during the MTTR period of the first processor when the first processor has already failed. 14 Load sharing operation Without detailed derivations, the result for MTBFD is MTBFD=(MTBF)2/2MTTR So, AD=MTBFD/(MTBFD+MTTR) =(MTBF)2/[(MTBF)2+2(MTTR)2] Therefore, UD=1-AD =2(MTTR)2/[(MTBF)2+2(MTTR)2] If MTBF>>MTTR, then, UD=2(MTTR)2/(MTBF)2 Ex: Given that MTBF=2000 hours & MTTR= 4 hours, calculate the unavailability for single & dual processor systems. Solution: U=2x10-3, 525 hours in 30 years UD=8x10-6, 2.1 hours in 30 years 15 Levels of control functions Considering the real time response requirements, the four important functions of a control subsystem may be grouped under three levels. The real time constraint necessitates a priority interrupt facility for processing in centralized control. If an event occurs when O&M function is being carried out by the control processor, the O&M processing has to be interrupted, the event processing taken up and completed , and then the O&M function processing resumed. 16 Interrupt processing Nesting of interrupts is necessary to suspend any low level function & to take up the processing of higher level functions. When an interrupt occurs, program execution is shifted to an appropriate service routine address in the memory through a branch operation. There are two methods of accomplishing this: Vectored interrupt- where the interrupting source supplies the branch address Non-vectored interrupt- where the branch address is fixed. The set of addresses is known as interrupt vector. 17
Download Complete (Ebook) Virtual Inertia Synthesis and Control by Thongchart Kerdphol, Fathin Saifur Rahman, Masayuki Watanabe, Yasunori Mitani ISBN 9783030579609, 9783030579616, 3030579603, 3030579611 PDF for All Chapters
Download Complete (Ebook) Virtual Inertia Synthesis and Control by Thongchart Kerdphol, Fathin Saifur Rahman, Masayuki Watanabe, Yasunori Mitani ISBN 9783030579609, 9783030579616, 3030579603, 3030579611 PDF for All Chapters