1089-7798 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2015.2495212, IEEE Communications Letters IEEE COMMUNICATIONS LETTERS, VOL. 14, NO. 8, AUGUST 2015 1 Handoff Decision Using a Kalman Filter and Fuzzy Logic in Heterogeneous Wireless Networks Iwan Kustiawan, Student Member, IEEE, and Kuang-Hui Chi, Senior Member, IEEE Abstract—Handoff decision is key to service continuity when a user roams about. We propose a Kalman filtering and fuzzy logic approach to reducing handoff initiations, so as to support seamless communication. Kalman filtering is used as a channel equalization technique in channel propagation model, while a Mamdani fuzzy logic is applied subsequently for handoff decision. Several metrics such as received signal strength (RSS), data rate, velocity of mobile terminal, and traffic load are considered as criteria to initiate handoff from WLAN to the cellular network. Simulation results show that the proposed method reduces handoff initiations effectively. Index Terms—Vertical handoff decision, Kalman filter, fuzzy logic, heterogeneous wireless networks. I. I NTRODUCTION N EXT generation wireless networks have dramatically raised the development of a diverse range of high speed multimedia services that lend themselves to ubiquitous com- puting. When a user roams about, handoffs among different radio access technologies occur frequently. Handoff decision is to determine the best prospective access network and decide at any given time whether or not to carry out handoff. In particular, handoff involving switching among different types of wireless network is referred to as vertical handoff. A smart connectivity engine should be designed to provide the best interface for any given application. A number of schemes have been proposed for vertical handoff decision in the literature. An advanced algorithm has been proposed to trigger handoff between 3G and WLANs using a cascaded exponential moving average filter [1]. This method has been evaluated with another filtering approach using a Kalman filter and both of methods showed good improvement in avoiding unnecessary handoff [2]. However, only received signal strength (RSS) alone was examined as a metric for handoff decision in these schemes. So, handoff is still susceptible to signal variation to some extent, leading to ping-pong effects, making switching among different networks back and forth happen repeatedly. Vertical handoff can be approached with fuzzy logic [3],[4]. These approaches consider various input parameters such as RSS, available network bandwidth, monetary cost, and Manuscript received June 1, 2015; accepted October 15, 2015. The associate editor coordinating the review of this letter and approving it for publication was Dr. M. Imran. I. Kustiawan is with the Graduate School of Engineering Science and Technology, National Yunlin University of Science and Technology, Yunlin 64002, Taiwan. E-mail: d10210039@yuntech.edu.tw. K-H. Chi is with the Department of Electrical Engineering, National Yunlin University of Science and Technology, Yunlin 64002, Taiwan. E-mail: chikh@yuntech.edu.tw. user preference as handoff decision criteria. However, these schemes operate using noise-inclusive RSS. As noise may interfere with the decision engine significantly, the ping-pong effect remains likely. One main contribution of this study lies in combining Kalman filtering and fuzzy logic techniques moderately to our purpose. Our treatment serves as an effective alternative deci- sion engine to support seamless communication by reducing handoff initiations in heterogeneous wireless networks. We still use RSS as a main metric to trigger handoff and then the filtered RSSs are combined with other metrics — data rate, velocity, and traffic load — as multicriteria for resolving a single output, namely handoff factor. For fuzzy logic, we adopt Mamdani fuzzy logic that controls signals. The Kalman filter is used here because of its salient feature and strength of providing quality estimates in case of Gaussian and linear models. The Kalman filter can be viewed as an es- timator with smallest possible error covariance which satisfies certain assumptions. However, if the system is nonlinear, we should consider other filtering techniques like particle filter to obtain better accurate estimates. We use fuzzy logic because of its unique characteristics, that is, capability to reason precisely under imprecise information. We use Mamdani fuzzy logic because of several reasons, including intuitive, widespread acceptance and convenience to human instinct. We remark that our scheme belongs to the combinational al- gorithm classified in [5]. Indeed, our scheme is complementary to previous studies in that our design can enhance or strengthen their operational efficiency, as shall be clarified below. II. HANDOFF DECISION A. Path loss model for channel modelling We use a log-linear path loss channel propagation model with shadow fading within WLAN given by [6]. The RSS is expressed in dBm as follows: RSS ( d ) = P T - L - 10 βlog 10 ( d ) + ϵ (1) where P T is the transmitted power, L is a constant power loss, β is the path loss exponent (with typical value of 2 to 4), d represents the distance between the mobile terminal (MT) and the Access Point (AP). ϵ is a zero-mean Gaussian random variable with standard deviation σ (with typical value of 6 to 12 dB) that represents the statistical variation in RSS ( d ) caused by shadowing.