Mobile Computing Systems Design Issues

"The technical design of mobile computing solutions offers unique challenges to systems professionals. Mobile users have unpredictable usage patterns, and ergonomic considerations in mobile office are formidable. Poor technical design and/or a disregard for ergonomics may ultimately lead to the failure of a mobile computing project." — Chander Dhawan - The Editor

We suggest that mobile computing professionals should analyze the following systems design issues:

1. Technical Design Issues

1.1 Network Design

1.1.1 Wireless LAN design issues

• How many mobile users in total will use the wireless LAN? How many will be active during the peak period?
• Which LAN applications will they be accessing? Remember that wireless LANs operate at much slower speeds than wired LANs. Will the slower speed be acceptable to the intended users?
• Is a notebook with a wireless NIC going to be the primary end user device? The use of notebooks as primary computers is being increasingly mandated even in fixed-location offices.
• In which areas of a building or campus will users be roaming?
• How many access points will be needed?
• Where should access points be located ?
• What is the access-point range?
• What impact will construction materials like steel frames used in walls and ceilings have on signal penetration?
• Which of the two main technologies is preferable — spread spectrum or frequency hopping?
• Will there be radio frequency interference from any other devices in the office, factory, or campus?

1.1.2 Wide Area Radio Network Design Issues

• Should a private radio network be built, or can a public shared radio network be used?
• Which radio network technology among those that are available now and will be available in the future, is most appropriate for the suite of applications that will be used?
• How are user application-usage profiles matched to a given network’s capability?
• Should RNA technology be integrated with a radio network infrastructure?
• How is good coverage and a minimum number of dead spots ensured?
• Should a distributed wireless network design with several MCSSs like the one suggested in Chapter 11 be used?
• Which MCSS should be selected?
• Should agent-based application development tools like Oracle Mobile Agents be used?
• How do various network design options influence the way logical networks will be managed?
• Is network outsourcing a realistic business option to pursue? If it is, does the proposed network service provider (NSP) have the necessary capacity and capability?

1.2 Capacity planning and response time calculations
A mobile computing application transaction traverses many hardware and software components before it reaches the destination server — and has to cover the same path again in reverse to complete the trip. There are many physical links (hops), wireless and wired line, between the end user’s client application software and the information server. There are also several pieces of software involved, many of which featuring queuing (i.e., they are asynchronous).

Thus, there are complex rules for scheduling priorities on a network. This makes it extremely difficult to build a mathematical model to estimate response times — and therefore to plan reliable network capacities in advance. Network providers do give some estimates, using either complex queuing models or rule-of-thumb calculations based on the experience of other customers.

1.3 Data compression considerations
Wireless network bandwidth is scarce and expensive. Every possible technique should be used to get the utmost service out of this bandwidth. Compression of data is one such technique.

The most common place to compress data is in the modem.  But it is a good idea to go beyond modem hardware in reducing the amount of traffic on wireless networks. Intelligence can be built into client application programs so that short message codes can be used to indicate common data occurrences. Forms and screens should be stored on the mobile computer.

1.4 System availability design
Typically, wireless networks have redundancies built into them and network service providers guarantee an extremely high network up-time percentage. Generally, base-station hardware and network controllers have hardware redundancy and message switches are typically built on fault-tolerant platforms such as Tandem and Stratus. However, nothing should be taken for granted and public shared network providers should be asked for details of their redundancies. Similar redundancies should also be built into private networks.

Another vital component that must have redundancy built into it is the MCSS. Find out if the method of switch-over from failing components to standby components is automatic or manual. If it is manual, is the station staffed, and how long does a switch-over take?

1.5 Security issues
Securing information from unauthorized access is a major problem for any network , especially for wireless networks, since radio signals travel through the open atmosphere where they can be intercepted by individuals who are constantly on the move — and therefore difficult to track down.

Here are some examples common security breaches:

• networks by criminal elements.
• Interception of credit card authorizations over wireless networks.
• Physical breach of security at unmanned base stations or other communications centers.

Therefore security design must be given careful consideration. Techniques such as on-the-air encryption (which is inherent in PCS networks) and firewalls must be used.

2. Ergonomic and Logistics Design
The designer should evaluate the following factors:

• Form factor of end-user devices
• Battery life
• Input method - keyboard, pen, touch or voice
• Ruggedness
• Health and safety issues
• Portable or fixed