Title: CASE STUDY OF END USER CONTROL: RONNY GOES TO TOKYO

This example gives a view of a possible application of AmI and the tension between end-user control and AmI environment control. The interaction between data subject and the environment is supported by a Personal Digital Assistant (PDA) that facilitates privacy and identity management by the end-user; Location Based Services (LBS); wireless communication, and a database containing addresses, opening hours and menus of restaurants. 

Ronny stays - for the first time - in Tokyo, to attend a workshop. He invites a colleague from Croatia to have dinner after the workshop. He needs to take a quick decision where to go and needs to know the restaurants that meet his preferences. Instead of wondering around without knowing where to go, he uses his PDA that contains his preferences for restaurants (see ).

His personal profile contains the following preferences: anonymity, classical music, diet food, local food, medium prices. The PDA sends a request to a database, containing a list of restaurants, with the opening hours and the characteristics of the restaurants.  

Figure :  Ronny’s PDA - communicating with restaurants in Tokyo.

The first restaurant, a steak house, plays rock music, has no diet menu and is medium- priced. The second restaurant is a local fish restaurant which offers diet menus and plays classical music. The prices are medium. The third restaurant is a fast-food chain that does offer diet menus and contains low prices. All restaurants are open at the day of request. The database receives the request, takes into account the opening hours of the restaurants and matches the restaurants’ data with Ronny’s preferences. Consequently, the contact address and location of ‘Fish Resto’ is displayed on the PDA of Ronny. He can make an immediate reservation. The preferences on his PDA form a profile that is communicated to his AmI environment. The restaurants themselves also use profiles. All persons that meet the requirements of such a profile are allowed to display the restaurant on the PDA. The restaurants that created a profile (data set) for clients that are willing to pay high prices, are not interested in Ronny who chooses medium priced restaurants.

In the case described so far, profiles are used that have nothing to do with profiling technologies. We are dealing with pre-defined input from the end-user and pre-defined input from the service provider. One could of course claim that this is not an example of ambient intelligence, but rather of ubiquitous computing. The environment is not learning, but just following predefined input. 

We can introduce software intelligence into this example by allowing the PDA and/or the devices of the service provider to record and process the actual behaviour of Ronny: how often he goes to which type of restaurant; which wines he chooses; whether he takes an entrée, a dessert; how far he likes to walk or travel from his hotel; whether he likes to meet new people, share dinner with other congress- participants or rather eat alone; how he takes his after dinner coffee; what time he gets home. This and other information may not be relevant immediately but could be built into a complex profile, stored on his PDA, accessible at different levels by different types of service providers. It could mean that fewer restaurants are displayed because the profile will be fine- tuned to a lot of different factors, matching with greater precision the offers of neighbouring restaurants. It could even be that those restaurants that do not present recently updated profiles with detailed information are discarded right away. (Of course Ronny may eventually come to live a rather boring kind of life, walking around in a world that satisfies his preferences - little chance of surprise.) To maintain end-user control all data collected and profiles constructed by the intelligent devices in the AmI environment should be stored on Ronny’s PDA - with the guarantee that they will not be stored with the service provider. The interesting question would be whether the same level of intelligence can be reached if only the PDA of the end user is able to construct profiles.