ements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripheralsThe behaviour of wireless ad-hoc networks is strongly determined by physical limitations of radio communication. Integrating the underlying physical laws into a simulator increases complexity, so that either simplifications are used or the scope of the simulator has to be reduced to a radio-layer simulator. While simulation studies are useful, they should be complemented with real world experiments in order to reveal phenomena the simulation cannot cover. For example, doing real world experiments with IEEE 802.11b based wireless ad hoc networks, Lundgren et al. discovered the existence of socalled communication gray zones that previous simulations were not able to detect.

For example, in NS-2 the standard way to model node delays is by using a timer and performing some actions on the expiration of the timer. In our testbed, the capabilities of the hardware are limited and therefore tasks such as packet processing requires much more time than in traditional routers. Real world experiments inherently include these increased processing costs. For these reasons, we decided to build a testbed using fully-understood hard- and software that allows to perform measurements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripherals

er simulator. While simulation studies are useful, they should be complemented with real world experiments in order to reveal phenomena the simulation cannot cover. For example, doing real world experiments with IEEE 802.11b based wireless ad hoc networks, Lundgren et al. discovered the existence of socalled communication gray zones that previous simulations were not able to detect.

For example, in NS-2 the standard way to model node delays is by using a timer and performing some actions on the expiration of the timer. In our testbed, the capabilities of the hardware are limited and therefore tasks such as packet processing requires much more time than in traditional routers. Real world experiments inherently include these increased processing costs. For these reasons, we decided to build a testbed using fully-understood hard- and software that allows to perform measurements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripherals

e of socalled communication gray zones that previous simulations were not able to detect.

For example, in NS-2 the standard way to model node delays is by using a timer and performing some actions on the expiration of the timer. In our testbed, the capabilities of the hardware are limited and therefore tasks such as packet processing requires much more time than in traditional routers. Real world experiments inherently include these increased processing costs. For these reasons, we decided to build a testbed using fully-understood hard- and software that allows to perform measurements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripherals

and therefore tasks such as packet processing requires much more time than in traditional routers. Real world experiments inherently include these increased processing costs. For these reasons, we decided to build a testbed using fully-understood hard- and software that allows to perform measurements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripheralsements and to evaluate scenarios covering all layers beginning at the radio-layer and embracing all layers up to HTTP/TCP/IP and the application space. The hardware is an embedded device we call Embedded Wireless Module (EWM) consisting of a core controller (Motorola 68HC912) and I/O peripherals as well as different wireless and wired network connectors. The device is, for example, equipped with Bluetooth and a 433 MHz RF module. In this paper, we demonstrate some experiments we have performed with this hardware. We present some general observations as well as some real world experiments covering different application domains such as ad-hoc gaming and home automation. While vertical handover times between Bluetooth and the 433 MHz RF modules are in the order of milliseconds, our measurements show that the handover time for a Bluetooth slave between two different piconets is too slow to enable highly interactive multi-user games in an ad-hoc gaming infrastructure we present in this paper.

The presented home automation scenario evaluates the efficiency of service usage in ad-hoc networks. While demonstrating the feasibility of this approach further optimizations of some of the steps are necessary. The remainder of this paper is outlined as follows: In the next section, we describe the hard- and software of this testbed. In Section 3, we discuss general observations found in our experiments and in Section 4 we present selected experiments. Finally we com