Seppo Hätönen defends his PhD thesis on Seamless Programmable Multiconnectivity
On Friday the 30th of September 2022, M.Sc. Seppo Hätönen defends his doctoral thesis on Seamless Programmable Multiconnectivity. The thesis is related to research done in the Department of Computer Science and in the Content-centric Structures and Networking group.

M.Sc. Seppo Hätönen defends his doctoral thesis Seamless Programmable Multiconnectivity on Friday the 30th of September 2022 at 13 o'clock in the University of Helsinki Exactum building, Auditorium B123 (Pietari Kalmin katu 5, 1st floor). His opponent is Professor  Michael Welzl (University of Oslo, Norway) and custos Professor Sasu Tarkoma University of Helsinki). The defence will be held in English. It is possible to follow the defence as a live stream at

The thesis of Seppo Hätönen is a part of research done in the Department of Computer Science and in the Content-centric Structures and Networking group at the University of Helsinki. His supervisors have been Professor Sasu Tarkoma and University Researcher Ashwin Rao (University of Helsinki).

Seamless Programmable Multiconnectivity

Our devices have become accustomed to being always connected to the Internet. Our devices from handheld devices, such as smartphones and tablets, to our laptops and even desktop PCs are capable of using both wired and wireless networks, ranging from mobile networks such as 5G or 6G in the future to Wi-Fi, Bluetooth, and Ethernet. The applications running on the devices can use different transport protocols from traditional TCP and UDP to state-of-the-art protocols such as QUIC. However, most of our applications still use TCP, UDP, and other protocols in a similar way as they were originally designed in the 1980s, four decades ago. The transport connections are a single path from the source to the destination, using the end-to-end principle without taking advantage of the multiple available transports.

Over the years, there have been a lot of studies on both multihoming and multipath protocols, i.e., allowing transports to use multiple paths and interfaces to the destination. Using these would allow better mobility and more efficient use of available transports. However, Internet ossification has hindered their deployment. One of the main reasons for the ossification is the IPv4 Network Address Translation (NAT) introduced in 1993, which allowed whole networks to be hosted behind a single public IP address. Unfortunately, how this many-to-one translation should be done was not standardized thoroughly, allowing vendors to implement their own versions of NAT. While breaking the end-to-end principle, the different versions of NATs also behave unpredictably when encountering other transport protocols than the traditional TCP and UDP, from forwarding packets without translating the packet headers to even discarding the packets that they do not recognize. Similarly, in the context of multiconnectivity, NATs and other middleboxes such as firewalls and load balancers likely prevent connection establishment for multipath protocols unless they are specially designed to support that particular protocol.

One promising avenue for solving these issues is Software-Defined Networking (SDN). SDN allows the forwarding elements of the network to remain relatively simple by separating the data plane from the control plane. In SDN, the control plane is realized through SDN controllers, which control how traffic is forwarded by the data plane. This allows controllers to have full control over the traffic inside the network, thus granting fine-grained control of the connections and allowing faster deployment of new protocols. Unfortunately, SDN-capable network elements are still rare in Small Office / Home Office (SOHO) networks, as legacy forwarding elements that do not support SDN can support the majority of contemporary protocols. The most glaring example is the Wi-Fi networks, where the Access Points (AP) typically do not support SDN, and allow traffic to flow between clients without the control of the SDN controllers.

In this thesis, we provide a background on why multiconnectivity is still hard, even though there have been decades worth of research on solving it. We also demonstrate how the same devices that made multiconnectivity hard can be used to bring SDN-based traffic control to wireless and SOHO networks. We also explore how this SDN-based traffic control can be leveraged for building a network orchestrator for controlling and managing networks consisting of heterogeneous devices and their controllers. With the insights provided by the legacy devices and programmable networks, we demonstrate two different methods for providing multiconnectivity; one using network-driven programmability, and one using a userspace library, that brings different multihoming and multipathing methods under one roof.

Avail­ab­il­ity of the dis­ser­ta­tion

An electronic version of the doctoral dissertation is available on the e-thesis site of the University of Helsinki at

Printed copies will be available on request from Seppo Hätönen: