Transparent IP Proxy for Tactical Ad hoc Networks Helder Marques, Jérémie Leguay, Hicham Khalifé, Vania Conan, Damien Lavaux Thales Communications & Security 4 rue des Louvresses, 92230 Gennevilliers, France Email: firstname.name@thalesgroup.com Abstract—This paper presents an adaptive and controllable framework to optimize the transport of IP packets in military MANETs. The HBH (Hop-By-Hop) protocol that we propose uses a combination of hop-by-hop reliability and congestion mitigation mechanisms. Its major advantage is to run any standard IP application unmodified on top of the network. Moreover it is able to improve flow performance in terms of latency or goodput. HBH is tunable and provides a target level of hop-by-hop reliability. It uses a transmission window and selective acknowledgements to improve bandwidth usage. We present the HBH protocol components, mechanisms and parameters. We carried out an implementation of HBH very close to a Linux implementation and validated it in the NS3 simulator. Finally, we demonstrate how HBH can improve TCP performance on a four-node chain topology by providing the right amount of additional reliability that the end-to-end TCP connection requires to overcome lossy conditions. I. I NTRODUCTION Mobile ad hoc networks (MANET) are dynamic networks composed of mobile devices connected wirelessly. Each node can move freely and communicate with the other nodes in radio range. Multiple radio hops may be necessary before a message reaches its destination. This type of architecture has hence the advantage of being very flexible and infrastructure- less, making it robust and useful for natural disaster recovery or military purposes. The next generation of military MANETs will support new forms of operational engagement, such as Network Centric Warfare [13]. Massive transformation pro- grams are following this path in the US [4] and in Europe [9]. Products and waveforms such as FlexNet [12], Falcon III/AN [3], or ESSOR [2], are either under development or starting to be deployed. Battlefield Management Systems (BMS) [1] are the most common type of applications that a tactical network would run, supporting message exchanges and distributing orders. As Fig. 1 shows, these applications may either use dedicated messaging interfaces to the radio through the SMTP or SOAP protocols, or use a standard IP interface. Most of the radios developed so far have been thus offering vertically integrated messaging interface. However, this reduces the interoperability with applications and other networks. On the other hand, IP and its extensions for group communications (multicast or Xcast) offer a standard mean to address military devices. In the latter case, a BMS or any data service needs to rely on standard transport protocols (TCP or UDP) that have shown major drawbacks in MANETs in general. The frequent link failures due to mobility, packet losses due to interferences Fig. 1: Connecting applications to tactical MANETs or jamming, and network capacity variations due to fluctuat- ing radio conditions, degrade considerably TCP performance, because of its end-to-end reliability and congestion control principle, whereas UDP lacks of any reliability control loop that can compensate lossy network conditions. To address these issues, this paper presents an adaptive and controllable framework, called HBH (Hop-By-Hop), to optimize the transport of IP packets in military MANETs. This solution aims at improving flow performance in terms of latency or goodput while being compatible with standard IP implementations to run unmodified applications on top of the network. Most of IP applications are using the TCP or UDP transport protocols. In practice, IP transport protocols fall into two extreme behaviors which are not flexible enough for tactical MANETs. On one hand, TCP offers congestion control and reliability mechanisms but has been designed for low latency and low packet loss networks. Its ACK based mechanism for full reliability makes it behave very poorly in lossy situations. On the other hand, UDP protocol does not use any feedback mechanism and is thus unable to ensure any type of control (i.e. reliability, congestion, fairness, etc...). HBH works transparently at the IP level in a hop-by-hop fashion to complement transport layer mechanisms that are end-to- end by definition. It provides a controllable level of hop-by- hop reliability and uses a transmission window and selective acknowledgements to improve bandwidth usage. The level of reliability can be controlled and chosen according to the nature of each IP flows. HBH also includes a hop-by-hop congestion mechanism to dynamically contain congestions outside of the