Nonparametric Bayesian Prediction of Primary Users' Air Traffics in Cognitive Radio Networks

Nonparametric Bayesian Prediction of Primary Users' Air Traffics in Cognitive Radio Networks

Ju Bin Song, Zhu Han
DOI: 10.4018/978-1-4666-6571-2.ch029
(Individual Chapters)
No Current Special Offers


In cognitive radio networks a secondary user needs to estimate the primary users' air traffic patterns so as to optimize its transmission strategy. In this chapter, the authors describe a nonparametric Bayesian method for identifying traffic applications, since the traffic applications have their own distinctive air traffic patterns. In the proposed algorithm, the collapsed Gibbs sampler is applied to cluster the air traffic applications using the infinite Gaussian mixture model over the feature space of the packet length, the packet inter-arrival time, and the variance of packet lengths. The authors analyze the effectiveness of their proposed technique by extensive simulation using the measured data obtained from the WiMax networks.
Chapter Preview


Due to non-cooperative nature and limitation of signaling in cognitive radio networks, it is difficult for secondary users (unlicensed users) to be aware of the activities of primary users. On the other hand, it is very important for the secondary users to be aware of primary user’s activity (such as air traffic patterns) in order to efficiently utilize primary user channel and prevent primary user from any harmful interference caused by secondary users. Channel occupancy statistics of primary users have high impacts on the performance of cognitive radio networks. For example, a secondary user can cluster the air traffic patterns of primary users, consequently, the secondary users can utilize the air traffic patterns to optimize their transmission strategy and save the unnecessary sensing overhead. Therefore, it is of significant importance for secondary users to detect the primary users’ air traffic patterns.

The rapid proliferation of Peer-to-Peer (P2P) and Skype applications have enriched the traffic over the Internet. The P2P traffic is using more than 50% of the total traffic on the Internet (Mellia, Pescape, & Salgarelli, 2009). Different types of applications show different behaviors in terms of features that have distinguishing characteristics (Bonfiglio, Mellia, Meo, Rossi, & Tofanelli, 2007; Choi & Hossain, 2011; Gu, Zhang, & Huang, 2011; Nguyen & Armitage, 2008). Features for different payloads can be packet (or frame) length (observed from packets’ TCP header), connection time (from SYN and FIN flags from the packet header), packet inter-arrival time (time between two consecutive packets of same application), the variance in packet lengths, and many others that are specific to different applications. Existing solutions for the identification of traffic payload falls into three categories. 1) Port based, this traffic payload identification, (Erman, Mahanti, Arlitt, Cohen, & Williamson, 2007), is not valid now anymore due to the fact that many applications use dynamic ports (like P2P) and try to disguise themselves by using known port numbers. 2) Payload based techniques that rely on deep packet inspection, (Haffner, Sen, Spatscheck, & Wang, 2005; Moore & Papagiannaki, 2005; Zander, Nguyen, & Armitage), that matches pre-defined signatures of applications. This technique performs well if the payload is not encrypted. Due to following reasons we cannot rely on payload based techniques, a) Encryptions of payloads prevent us to inspect the packet, b) Signatures of applications may change with time, different versions have different signatures (MSN2009 and MSN2011), c) Obfuscated data can lead to serious problems, Skype application uses data obfuscation (Bongfiglio, Mellia, Meo, Rossi, & Tofanelli, 2007). 3) Heuristic approaches are not accurate for classifying traffic applications (Bongfiglio, Mellia, Meo, Rossi, & Tofanelli, 2007; Mellia, Pescape, & Salgarelli, 2009).

Key Terms in this Chapter

Bayesian Nonparametric Clustering of Traffic Pattern: A nonparametric Bayesian approach for clustering traffic applications, where the nonparametric means the number of hidden traffic applications are unknown. The inference model estimates traffic applications for each observed primary user’s feature point given the feature space, the model parameters, and the hyper-parameters.

Traffic Pattern: An air-traffic pattern between sender and receiver of primary users. An air traffic pattern has a unique distribution of packet inter-arrival time and length depending on application.

Cognitive Radio: An intelligent radio that can be programmed and configured dynamically. Its transceiver is designed to use the best wireless channels in its vicinity. Such a radio automatically detects available channels in wireless spectrum, then accordingly changes its transmission or reception parameters to allow more concurrent wireless communications in a given spectrum band at one location. This process is a form of dynamic spectrum management.

Short Time Opportunity (STO): An inter-arrival time between primary user’s traffic applications when primary user is on.

Secondary User: An unlicensed user with cognitive radio function.

Primary User: A licensed user.

Traffic Pattern Identification: Clustering air traffic pattern which depends on application.

Long Time Opportunity (LTO): A long idle period of primary user when it doesn’t use any spectrum.

Complete Chapter List

Search this Book: