Agent Technology

Agents are software entities that can access remote databases—software objects that
can transport themselves from (electronic) place to (electronic) place.
The following are typical agent capabilities:
 An agent can travel to meet and then interact (for example, negotiate) with
another agent.
 Agents can be given a go instruction with a ticket that confers an authority to
meet, refer, negotiate, buy, sell, or barter.
 On arrival, the agent presents a petition (for example, the requirement, how
long the agent can wait).
 Agents can gather, organize, analyze, and modify information.
 Agents can have limits: time (for example, a 5-minute agent), size (a 1-kbyte
agent), or spending (a $1 million agent).
 Because it has authority, an agent can negotiate locally without referral back to
its master, which means it is well suited to being disconnected from a network.
 You can also send an agent instructions and messages, such as go to sleep,
wake up, or welcome back home.
The problem with agents is that they are analogous to computer viruses—in that,
they work in a similar way though without malicious intent. Agents therefore need
very good consistent authentication. Given that you are effectively allowing software
devices to spend money on your behalf, then it is important to have consistent rule
management. There is an implicit need to establish and maintain trust between people
and machinery. The need to establish trust has to involve mutually suspicious
machines (machines that must prove their identity to one another) and mutually suspicious
agents (agents that must prove their identity and authority to buy or sell to
each other).
It is difficult to establish a consistent and stable trust hierarchy, and to date this has
prevented the wide scale deployment of agent technologies. The SIM/USIM-enabled
smart card is arguably the pivotal software/hardware component needed to make
agent technology realizable on a given basis. Unfortunately, to date, smart card penetration
in the United States has been significantly slower than the rest of the world, and
this has hindered mass market adoption of agent-based services in digital cellular networks.
The gradual integration of the SIM/USIM (a work item in 3GPP1 and 3GPP2)
will help bridge the gap between the United States and the RoW (rest of the world)
agent technology platforms.

Summary
As we will see in our next 10 chapters, it is increasingly important to qualify how handset
hardware and software impacts on network hardware and software topology.
Specifically, we must qualify how the value generated by handset hardware and software
form factor and functionality must be preserved as the product (authenticated
and encrypted rich media, parallel application streaming, e-commerce, m-commerce,
and so forth) is moved into and through the core network.