Biblio

While we cannot question the high performance capabilities of the kernel bypass approach in the network functions world, we recognize that the Linux kernel provides a rich ecosystem with an efficient resource management and an effective resource sharing ability that cannot be ignored. In this work we argue that by mixing kernel-bypass and in kernel processing can benefit applications and network function middleboxes. We leverage a high-performance user space TCP stack and recent additions to the Linux kernel to propose a hybrid approach (kernel-user space) to accelerate SDN/NFV deployments leveraging services of the reliable transport layer (i.e., stateful middleboxes, Layer 7 network functions and applications). Our results show that this approach enables highperformance, high CPU efficiency, and enhanced integration with the kernel ecosystem. We build our solution by extending mTCP which is the basis of some state-of-the-art L4-L7 NFV frameworks. By having more efficient CPU usage, NFV applications can have more CPU cycles available to run the network functions and applications logic. We show that for a CPU intense workload, mTCP/AF\_XDP can have up to 64% more throughput than the previous implementation. We also show that by receiving cooperation from the kernel, mTCP/AF\_XDP enables the creation of protection mechanisms for mTCP. We create a simulated DDoS attack and show that mTCP/AF\_XDP can maintain up to 287% more throughput than the unprotected system during the attack.

The growth of malware poses a major threat to internet users, governments, and businesses around the world. One of the major types of malware, ransomware, encrypts a user's sensitive information and only returns the original files to the user after a ransom is paid. As malware developers shift the delivery of their product from HTTP to HTTPS to protect themselves from payload inspection, we can no longer rely on deep packet inspection to extract features for malware identification. Toward this goal, we propose a solution leveraging a recent trend in networking hardware, that is programmable forwarding engines (PFEs). PFEs allow collection of per-packet, network monitoring data at high rates. We use this data to monitor the network traffic between an infected computer and the command and control (C&C) server. We extract high-level flow features from this traffic and use this data for ransomware classification. We write a stream processor and use a random forest, binary classifier to utilizes these rich flow records in fingerprinting malicious, network activity without the requirement of deep packet inspection. Our classification model achieves a detection rate in excess of 0.86, while maintaining a false negative rate under 0.11. Our results suggest that a flow-based fingerprinting method is feasible and accurate enough to catch ransomware before encryption.

It has been shown that adversaries can craft example inputs to neural networks which are similar to legitimate inputs but have been created to purposely cause the neural network to misclassify the input. These adversarial examples are crafted, for example, by calculating gradients of a carefully defined loss function with respect to the input. As a countermeasure, some researchers have tried to design robust models by blocking or obfuscating gradients, even in white-box settings. Another line of research proposes introducing a separate detector to attempt to detect adversarial examples. This approach also makes use of gradient obfuscation techniques, for example, to prevent the adversary from trying to fool the detector. In this paper, we introduce stochastic substitute training, a gray-box approach that can craft adversarial examples for defenses which obfuscate gradients. For those defenses that have tried to make models more robust, with our technique, an adversary can craft adversarial examples with no knowledge of the defense. For defenses that attempt to detect the adversarial examples, with our technique, an adversary only needs very limited information about the defense to craft adversarial examples. We demonstrate our technique by applying it against two defenses which make models more robust and two defenses which detect adversarial examples.

In this paper, we propose and implement CommunityGuard, a system which comprises of intelligent Guardian Nodes that learn and prevent malicious traffic from coming into and going out of a user's personal area network. In the CommunityGuard model, each Guardian Node tells others about emerging threats, blocking these threats for all users as soon as they begin. Furthermore, Guardian Nodes regularly update themselves with latest threat models to provide effective security against new and emerging threats. Our evaluation proves that CommunityGuard provides immunity against a range of incoming and outgoing attacks at all points of time with an acceptable impact on network performance. Oftentimes, the sources of DDoS attack traffic are personal devices that have been compromised without the owner's knowledge. We have modeled CommunityGuard to prevent such outgoing DDoS traffic on a wide scale which can hamstring the otherwise very frightening prospects of crippling DDoS attacks.

Software-defined Networking (SDN) enables advanced network applications by separating a network into a data plane that forwards packets and a control plane that computes and installs forwarding rules into the data plane. Many SDN applications rely on dynamic rule installation, where the control plane processes the first few packets of each traffic flow and then installs a dynamically computed rule into the data plane to forward the remaining packets. Control plane processing adds delay, as the switch must forward each packet and meta-information to a (often centralized) control server and wait for a response specifying how to handle the packet. The amount of delay the control plane imposes depends on its load, and the applications and protocols it runs. In this work, we develop a non- intrusive timing attack that exploits this property to learn about a SDN network's configuration. The attack analyzes the amount of delay added to timing pings that are specially crafted to invoke the control plane, while transmitting other packets that may invoke the control plane, depending on the network's configuration. We show, in a testbed with physical OpenFlow switches and controllers, that an attacker can probe the network at a low rate for short periods of time to learn a bevy of sensitive information about networks with \textbackslashtextgreater 99% accuracy, including host communication patterns, ACL entries, and network monitoring settings. We also implement and test a practical defense: a timeout proxy, which normalizes control plane delay by providing configurable default responses to control plane requests that take too long. The proxy can be deployed on unmodified OpenFlow switches. It reduced the attack accuracy to below 50% in experiments, and can be configured to have minimal impact on non-attack traffic.

In this paper, we study information leakage by control planes of Software Defined Networks. We find that the response time of an OpenFlow control plane depends on its workload, and we develop an inference attack that an adversary with control of a single host could use to learn about network configurations without needing to compromise any network infrastructure (i.e. switches or controller servers). We also demonstrate that our inference attack works on real OpenFlow hardware. To our knowledge, no previous work has evaluated OpenFlow inference attacks outside of simulation.