Methods for service discovery in Bluetooth scatternets Nils Agne Nordbotten a, * , Tor Skeie a , Niels D. Aakvaag b a Simula Research Laboratory, P.O. Box 134, NO-1325, Lysaker, Norway b ABB Corporate Research, P.O. Box 94, NO-1375, Billingstad, Norway Abstract This paper presents methods for service discovery in multi-hop Bluetooth ad hoc networks, so called scatternets. Two service discovery protocols based on filtering of service requests are proposed. Extensive simulation results are presented showing that the protocols significantly reduce network traffic. Reducing the network traffic is important as many Bluetooth devices have limited power sources and, therefore, benefit from keeping links idle in power saving modes. It is also explained how the proposed protocols can interact with reactive routing protocols and effectively assist route discovery. Finally, an implementation providing functionality for both searching and browsing for services is suggested, effectively extending the Bluetooth SDP to the scatternet. q 2004 Elsevier B.V. All rights reserved. Keywords: Bluetooth scatternets; Service discovery; Routing 1. Introduction Bluetooth [1] is a short-range wireless communication technology. The basic communication structure in Blue- tooth is the piconet, a master node connected with one to seven slaves, where communication is conducted in a master to slave fashion. A network consisting of two or more coupled piconets is called a scatternet. The creation of ad hoc IP-based networks is facilitated in Bluetooth by the Personal Area Networking (PAN) Profile [2]. This profile allows the master to forward IP-packets encapsulated by a protocol similar to the well-known Ethernet protocol, namely the Bluetooth Network Encapsu- lation Protocol (BNEP) [3], enabling communication between all nodes within a piconet. An extended future PAN specification is expected to specify ad hoc Bluetooth networks spanning multiple piconets, where communication can be conducted across piconet borders. Such a network enables communication between more than eight Bluetooth devices, the maximum number of active nodes in a Bluetooth piconet, and enables communication between nodes otherwise out of range by the use of multi-hop paths. Such ad hoc networks will be dynamic in nature and should function without pre-existing infrastructure. In networks like this a method for service discovery is needed to avoid excessive user configuration. The Bluetooth specification defines the Bluetooth Service Discovery Protocol (SDP), a service discovery protocol for use in this network technology. This protocol is based on connecting to a specific node and querying about its available services. When networks become larger, as they will when multiple piconet PANs become a reality, querying every single node in the network about a possibly unavailable service wastes both time and resources. Another way to achieve service discovery should therefore be used in scatternets. Other service discovery protocols exist, like the Service Location Protocol (SLP) [4], Jini [5] from Sun Micro- systems, and UPnP’s [6] Simple Service Discovery Protocol (SSDP). Common for these protocols is that they are designed with the traditional wired network in mind. These networks have quite different characteristics than Bluetooth scatternets, where bandwidth and power are limited resources, latency behaviour is likely to be quite different from traditional LANs, and the network topology is likely to change as nodes enter and leave the network. Some work has been done on service discovery in mobile ad hoc networks, and the subject is for example discussed in Ref. [7,8]. Bluetooth scatternets are different from these networks, however, because of their connection oriented nature. For this reason a new service discovery scheme is needed. A solution based on caching service discovery replies [9] has been suggested to reduce network traffic. However, such an approach does not eliminate broadcast queries flooding 0140-3664/$ - see front matter q 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.comcom.2004.01.013 Computer Communications 27 (2004) 1087–1096 www.elsevier.com/locate/comcom * Corresponding author. E-mail addresses: nilsno@simula.no (N.A. Nordbotten); tor.skeie@ simula.no (T. Skeie); niels.aakvaag@no.abb.com (N.D. Aakvaag).