**3. The bidding strategy framework**

As mentioned, this work is an extension of Anthony's work (Anthony, 2003) to tackle the problem of bidding in multiple auctions that employ varying auctions protocols. This section details the electronic marketplace simulation, the bidding strategies and the genetic algorithm implemented in the previous work.

#### **3.1 The electronic market place simulation**

The market simulation employed three different auction protocols, English, Vickrey and Dutch that run simultaneously in order to simulate the real auction environment. The market simulation is used in this work to evaluate the performance of the evolved bidding strategies. The following section explains how the market simulation works.

The marketplace simulator shown in Fig. 2 consists of concurrent running auctions that employ different protocols. These protocols are English, Dutch and Vickrey. All of these auctions have a known starting time and only English and Vickrey auctions have a known ending time. The bidding agent is given a deadline (*tmax*) by when it must obtain the desired item and it is told about the consumer's private valuation (*pr*) for this item. The agent must only buy an instance of the desired item.

The marketplace announces the current bid values and the current highest bids for English auctions and the current offers for Dutch auctions at each time step. At the end of a given auction, it determines the winning bid and announces the winner. This set of information is used by the agent when deciding in which auction to participate, at what value to bid and in which time to bid.

Fig. 2. The Marketplace Simulator
