Specify what you need, not what is available
Specify the size of the system
Don't just quote the number of users etc that System X is able to deliver. How many do you need? As a recommendation use the following formula:

What is the expected maximum occupancy of the site? (A)

What is the expected life time of the system? (B)

What is the turnover in occupants likely to be (1=low, 5=high)? (C)

Number of current users = (A) x 1.3

Number of users total = (A) x (B) x (C) / 5

Specify the number of access levels required.
As a general rule of thumb allow for at least 3 access levels for each tenant/department. Where specific times are required multiply this number by the number of time combinations. An alternative rule of thumb is to multiply the number of doors in the system by 4.

Specify the number of time zones required.
Allow one time zone per tenant/department.

Graphical overkill
Probably the biggest gimmick on the market for access control is the graphical interface. It is an expensive and unnecessary option for many installations. It is strongly recommended to have a graphical interface where there are a large number of doors with many operators. For systems with only one building manager, it is unnecessary.

Impractical ideas
When considering the operation of the system, consider if a feature is really viable. For example, anti-passback is occasionally requested in car parks. The feature is installed and turned on, but within a month it is turned off again because of the complaints. In this case, consider something more practical such as a car park counter where groups are assigned a fixed number of places.

Consider how the users will interact with the system
In particular, work out how people will leave the area. Options include:

• request to exit push button
• PIR detector to detect a person approaching the door. Ensure that the PIR does not trigger the strike. - switch in the handle. We strongly recommend that this method is not used. It results in too many failures as the handle ages.
• mechanical handle. The problem is that you can't differentiate between a person walking out and breaking in.

Don't compromise security
Site codes
Many systems do a separate site code and credential number check. If the site code is in its list, and the credential number is in its list then the user has access. This method provides a huge breach in security. For example, if card number 10, site code 10 is programmed in, and card number 20, site code 20 is added, then card 10, site code 20 and card 20, site code 10 are also automatically permitted. Under this method, the total number of users given access is the number of cards times the number of site codes. The only secure method is to store the site code and card number together and to test both together.

Credential uniqueness
Consider how easy it is to reproduce the credential. Keypads should only be used where price is a real issue and other security measures are backing up the system. PINs of at least six digits should be used. Wiegand provides only a limited number of site codes. It is possible to have other cards in a city that will have access to the site. Silicon keys provide a totally unique number that can not be reproduced.

Fully distributed system
The database must be local to all controllers. Too many systems rely upon a single master controller to do their security processing. This not only slows the system down, but also provides a single point of failure. If the main controller died, the system would either enter a degraded mode or stop completely.

Automatic download of changes
All changes must be automatically downloaded to the controllers. Systems that rely upon the operator to change programs or do a special action are slow and prone to error.

Give a second thought to the credential
There are now a wide range of credentials and readers on the market. Many of these are now obsolete but have a traditional market. Others are experimental technologies with an unknown future. For most applications we recommend either a proximity or silicon key system. We generally recommend that mag-swipe and keypad systems be avoided. Proximity cards provide the benefits of ease of use and photo ID. The cards are relatively inexpensive and the readers have dropped considerably in price. Their main drawbacks are the inconvenience of carrying and susceptibility to be left at home or lost. Silicon keys provide the benefits of being vandal resistant (both the key and reader) as well as being very easy to carry. The readers are very inexpensive but the keys are slightly more expensive than many of the card based systems. Magswipe cards have a large market where price is a major issue. They are susceptible to magnetic fields and wear out with repeated use. Keypads provide free credentials but with the problem of easy duplication (anyone can remember a PIN). They are popular where a large number of people need access to an area and security is not critical or where additional security measures have been made.

Consider the future
Wear and tear
The Silicon Key reader and keys have a life time warranty. They will handle the knocks and blows that happen every day and will with stand a fair amount of vandalism. For durability there is no better option. Magstripe are probably at the other end of the scale. It relies upon contact and the cards and readers will wear out with repeated use.

Beware of the sting
Many proprietary systems have hidden expenses down the line. The most common one is excessive charges on credentials. Don't be locked into an expensive reader/credential that is only available from one supplier.