SWZ  >  
Research Projects  >  
Securing Communications in Internet of Things (IoT) Environments

Securing Communications in Internet of Things (IoT) Environments

Establishing the connectivity and delivering information for self-configuring wireless nodes in resource-constrained Internet of Things (IoT) environments is prone to security holes.

IoT has been recently extensively investigated through the introduction of various innovative network infrastructure designs such as mobile ad hoc networks (MANET), delay-tolerant networks (DTN) and information-centric networks (ICN).

Although a couple of proposals have been proposed for securing these infrastructure designs, dealing with potential attacks against neighbor discovery and localization such as wormhole attacks is an issue which has received considerable attentions recently. Secure neighbor discovery (SND) and secure localization play a critical role in location-based services which is a primary in IoT environments.

Another security issue in IoT is to detect malicious or selfish nodes and reduce unnecessary traffic on the network as well as to end devices.

The main idea is to protect spam in the "content" level rather than the "communication level", which ICN provides a great opportunity:

Spam might incur great workload in IoT environments. Large amount of spam wastes already precious network bandwidth, affects the existing timely communication, and might also cause denial of service (DOS) in the network. Even the network in which neighbors are authenticated cannot avoid spam with high accuracy and low false positive. The Email system is one of the examples. People still receive huge amount of spam when every mail server is authenticated by the DNS. We believe that spam can only be solved in the content level rather than communication level.

In this part of the work, we are going to find a way to protect spam in IoT with the help from ICN. ICN might provide a new way to solve the long-standing spam problem. In ICN, data are forwarded according to the Content Name or Content Descriptor. In order to disseminate the spam to the victims, spammers usually need to give spam a Content Name or a series of Content Descriptors, which partially reveals the content of the spam. But such information can also help the network identify the redundancy of the information disseminated (one key feature of spam is high redundancy). That gives us a possible way to monitor the amount of original information in a series of data sent to the network and try to identify spam at the “content” level.

Therefore, one key requirement of this environment is the ability to ensure that neighborhood discovery is securely performed and the wormholes can be prevented. Another key requirement is the ability to control spam and ensure that suitable denial-of-service capabilities are also built into the IoT architecture from the very beginning.

With the help of the simulation center, we can study the behavior of spammers and get the statistical information about how a spam is disseminated throughout the IoT network. Using the results, we can then fine tune metrics and threshold of the spam classifier. The verification of the spam protector can also be partially done in the simulation environment. The overhead and the benefit for different spam protectors can also be studied and compared to further optimize our spam protector.

Motivation to secure and accurate localization

Location-based services in IoT applications, need to rely on location information of nodes. There are variety of applications for sensor and ad hoc networks in which performing node localization in a secure and correct way is critical. For example in healthcare applications where the wireless medical sensor nodes are used to monitor the patient, the sensor nodes need to obtain their location information to allow the medical staff to reach the patient in the case of any emergency.

Another application could be in vehicular networks. Vehicles might send information about car crashes or other dangerous situations. In this case we need to obtain an accurate and verified location information to ensure other cars receive the warning in time and for the correct spot.

Although GPS receivers are easily installed on vehicles, GPS may have some undesired problems such as not always being available or not being robust enough for some applications. Therefore, other localization techniques would be useful for vehicular networks in such situation.

Another application is police and fire-fighter missions in which action forces wear sensor nodes. In such applications, sensors thrown into buildings to explore the conditions need to deliver results with verified location information to ensure action forces don't run into a disaster. As a security issue, relying on localization information can be endangered by tampered anchor node announcements.

Neighbor discovery can be seen as an important requirement to many protocols such as routing and localization in wireless networks. In localization protocols the nodes usually must rely on the available location information of their neighboring anchor nodes who are aware of their own location (e.g. using GPS). A very subtle attack against reliable neighbor discovery in wireless networks is wormhole attack, in which the attacker eavesdrops the neighbor discovery messages in one area in the network and then tunnels and replays them to another typically far away area. Such an attack would make the nodes in the two areas to believe to be neighbors while they are actually not. 

In presence of a wormhole attack, when nodes rely on the location information of anchors incorrectly detected in their neighborhood to determine their coordinates, the localization protocol would be severely affected. In such a case the location announcement will be a false one (assumed by the receiver to have been issued by a correct neighbor while it is not) and therefore will input wrong location information to the localization protocol.

Involved Scientists

Prof. Dr. Dieter Hogrefe

Telematics Group
Institute of Computer Science
University of Göttingen

Goldschmidtstraße 7, 37077 Göttingen
E-Mail: hogrefe@informatik.uni-goettingen.de
Phone: +49 551 39-172001
Fax: +49 551 39-14403

Prof. Dr. Xiaoming Fu

Computer Networks Group
Institute of Computer Science
University of Göttingen

Room 3.108, Goldschmidtstraße 7, 37077 Göttingen
E-Mail: fu@cs.uni-goettingen.de
Phone: +49 551 39-172023
Fax: +49 551 39-14416

Publication list

2014

2013