Limiting the loss of information in KNXnet/IP on congestion conditions Salvatore Cavalieri ⇑ , Ferdinando Chiacchio University of Catania, Department of Electrical Electronic and Computer Engineering, Viale A. Doria 6, 95125 Catania, Italy article info Article history: Received 16 May 2013 Received in revised form 16 August 2014 Accepted 18 August 2014 Available online 27 August 2014 Keywords: KNXnet/IP Congestion control Forwarding policy Performance evaluation Petri net SAN abstract KNXnet/IP communication system allows integration of different KNX networks through IP medium by a particular device called KNXnet/IP Router. Due to the different transmission speeds among KNX networks and the IP medium, a control mechanism to prevent the con- gestion of KNXnet/IP Routers has been foreseen by the KNXnet/IP standard. In this paper, the authors analyse the performance of the KNXnet/IP congestion control mechanism in terms of its impact on the loss of information exchanged between KNX devices located in different KNX networks. The main goal is to point out suitable configurations of the congestion control mechanism capable, more than others, to limit the loss of information. Performance evaluation has been realised through simulation of a Petri Net model based on Stochastic Activity Network (SAN), capable of implementing the main features of the KNXnet/IP specifications. Ó 2014 Elsevier B.V. All rights reserved. 1. Introduction KNX is one of the most well-known and worldwide used standards for home and building automation [1–5]. It was defined several years ago and later it was integrated into IP environment through the definition of the KNXnet/ IP specification [6]. KNXnet/IP has quickly gained the attention of vendors and users; also researchers are inter- ested in several relevant issues, like performance [7–9] and security [10]. Integration of KNX with IP is based on the use of a par- ticular device called KNXnet/IP Router, which allows con- nection of several KNX networks to an IP medium used as backbone. Due to the very limited transmission speeds foreseen for certain types of KNX networks, the KNXnet/ IP Router may receive more messages from IP network than it is able to sell off; in this case all or parts of the incoming messages may be lost. That condition is relevant to congestion; in order to mitigate this issue, a mechanism of congestion control has been introduced in the KNXnet/IP standard [6]. Basically, the KNXnet/IP congestion control is based on a very simple stop-and-go strategy; when a KNXnet/IP Router detects an internal congestion condition, it sends in multicast a particular datagram, causing the stop of all transmissions over IP by the other KNXnet/IP Routers. Communications are resumed when the congestion condi- tion inside the KNXnet/IP Router has been surpassed. It is clear that the activation of KNXnet/IP congestion control mechanism may cause loss of information exchanged by control devices located in different KNX net- works and may affect the behaviour of the control applica- tions running on KNXnet/IP communication system. For example, let us consider a control scenario featuring the presence of a controller located in a KNX network which has to receive the values generated by several sensors http://dx.doi.org/10.1016/j.comnet.2014.08.012 1389-1286/Ó 2014 Elsevier B.V. All rights reserved. ⇑ Corresponding author. E-mail addresses: salvatore.cavalieri@dieei.unict.it (S. Cavalieri), chiacchio@dmi.unict.it (F. Chiacchio). Computer Networks 73 (2014) 154–172 Contents lists available at ScienceDirect Computer Networks journal homepage: www.elsevier.com/locate/comnet