The LOD Technical Journal: File #7 of 12

         (*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)
         Testing Operations Provisioning Administration System
                               (TOPAS)
                       LOD - Mystik Freak - LOD
         (*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)(*)


In order to perform Operations, Administration and Maintenance (OA&M) on
switched circuit and facility networks the TOPAS operating system (OS) has
been developed.  From the "core" of TOPAS the Transport Maintenance
Administration System (TMAS) was designed to assist in running the Facility
Maintenance and Administration Center (FMAC).

As the telephone network became more and more advanced the conduction of OA&M
became increasingly difficult. What's brought about this sophistication has
been the later versions of electromechanical switching systems, ISDN etc. In
order to keep up Artificial Intelligence (AI) ideas are being used as a basis
for TOPAS-ES. TOPAS-ES is designed as an Expert System (ES) replacement for
TOPAS to handle switch circuit operations. TOPAS-ES performs this circuit
maintenance using its AI to find and report on network difficulties. 

Network Maintenance
In the current 5ESS Switch maintenance is performed by TOPAS and the remote
measurement system (RMS-D3). Under 4ESS circuit maintenance system 1 (CMS-1)
is used. The purpose of RMS-DX is to allow testing on circuits terminating on
switches. The network is monitored as the transmission passes through the
XESS Switch, the multiplexer (MUX) and the line terminating equipment (LTE).

TOPAS and CMS-2 continually monitor the network's status and look for
deviations from normal operations and then print up trouble reports. Because
so many reported problems are transient or falsely reported as a problem,
further testing is done to determine real or "hard" problems. Through such
procedures as performing tests on one of more than a million scan points or
attempting to receive from one or two ends of the circuit. 

TOPAS uses two different machines with their own databases when processing:
Equipment Interface Tier (EIT) and the Network Support Tier (NST). 

EIT - An EIT contains a database that has physical information about a
Network Element (NE) machines.

NST - NST's databases are not interested in NE machines or in physical
properties and instead uses mathematical models. Even radical network changes
will have only minimal effects. Thus the combination of say fiber and copper
wiring on the same circuit or the merging of voice and data communications
has no great effect. NST can handle everything from basic trunking to complex
multipoint circuits. 

Both EIT and NST use Common Languages to communicate with each other. NST
will for example query NST about specific equipment, while EIT would query
NST about network changes. Since EIT and NST are both in the TOPAS core
interactions are quite simple. 

TMAS
TMAS followed TOPAS and in its design, developers reused almost half of
TOPAS's core. Since TOPAS and TMAS speak a common language cooperation
between the two is possible. Many report procedures are identical such as the
DS-1 facility alarms.

FMAC
TMAS is designed to run with the FMAC. By providing updated route databases,
alarm monitoring, detection of network faults etc. TMAS also helps
administrate by issuing trouble tickets, switch logs and sending out this
data to other personnel from the FMAC.    
 
Expert Systems (ES)
An ES is a system where the program and the knowledge used in decision making
are kept apart. The program contains a set of rules, containing what action
should be undertook depending on the situation. This is often referred to as
a "shell" that controls the activities of its host system (think of the UNIX
shell). 

ESs in Networks
The maintenance of complex networks is an ideal application for an ES. By
having the equivalent of the most capable repair mind on each switch. As all
the ESs are using a common knowledge base that has everything known about the
problem and the most effective way to solve it. Several other ESs have
predated TOPAS-ES such as ACE, NEMESYS and GTE's COMPASS. As any technical
worker will attest to, network operations are particularly troublesome as the
call carrying capacity must be maximized while trying to minimize the
congestion that results when traffic exceeds the call capacity of the
switching and transmission system.

TOPAS-ES
TOPAS-ES, is as the name indicates, is an ES version of TOPAS. It works with
both TOPAS and CMS-1 in the 4ESS and 5ESS environment. TOPAS-ES has a UNIX
routine for each of its three subsystems - knowledge base and inference
engine, communication and systems interface and user interface. The inference
engine used in TOPAS-ES is "forward chaining" or data driven as it is guided 
available data to fit prestated conditions to obtain an answer. If it used
backward chaining, it would search for data to obtain an answer. Forward
chaining is a more effective route to take when data is available and answers
to a question (using backward chaining) are unneeded or to slow. Generally,
forward chaining in network maintenance is preferred. For example, data
indicating that Joe Phreaker is blowing 2600 tones is of more use than
attempting to answer a question of "Where are all the foreign tones on the
circuit originating from?" To keep up with its immense chores of network
monitoring, testing and issuing trouble reports, gathering data and figuring
out answers TOPAS-ES runs each of its subsystems at the same time, working in
"real time" with the network.  

Distributed AI (DAI)
DAI is where multiple processes which normally act independently, co-operate
which one another. TOPAS-ES uses DAI to station one TOPAS-ES at one end of
the circuit and another on the other end or at the CO. This enables more
computing power to be levied at pinpointing the problem and makes for a
faster, more reliable system. TOPAS-ES can assume either a director or
responder mode. If TOPAS-ES is analyzing a faulty circuit it can request or
enlist another TOPAS-ES and place it in the responder mode to assist it. 

Expert System Trouble Analyzer (ESTA) 
This is one of TOPAS-ES's subsystems and performs the main operations of:
trouble ticket analysis and chronic history analysis (CHA). 

Trouble ticket analysis: Since few problems reported by TOPAS-ES are genuine
ones that require attention, ESTA narrows down the hard from the transient
problems. ESTA determines this mostly by ordering TOPAS-ES to wait and
perform further monitoring.

CHA: This exposes faults after repeated transient trouble indications. If the
problem persists for longer than X amount of time, with over Y indications of
trouble it will be labelled chronic. CHA is designed to pick up on problems
that have been passed off as transients and ignored. For example a problem
may exist during peak hours but will be passed off as a transient when
monitored during off-peak hours.

Expert System Trouble Sectionalizer (ESTS)
Once ESTA has determined a trouble to be hard it will pass along a "trouble
ticket" indicating such information as its duration, current condition and
whether its chronic or not. 

When ESTS has been handed a hard trouble it will "sectionalize" the indicated
area on the circuit. This is done by having technicians at each end examine
points on the circuit and performing other tests. ESTS is based on the best
sectionalization techniques, being an ES. An ESTS sectionalization strategy
would work like this: 2600 tones are being heard on the network, circuits are
all in normal condition, 2600s are not in internal use and have been labelled
as unauthorized, foreign sounds so ESTS would deduce that someone is trying
to bluebox. 

ESTS has a wide list of strategies to try depending on the situation. The
most likely to succeed strategies will be attempted first and if this fails
all of its strategies will be tried in order of success probability. Once the
fault has been pinpointed the relevant repair crew/station will be notified
along with a description of the fault.

---------------------------------------------------------------------------

