Optimal Handoff And Bandwidth Aggregation In Heterogeneous Wireless Networks

With the advent of the era of mobile Internet, wireless broadbandtechnology in recent years has made rapid development. The Internet access isbecoming ubiquitous with the rapidly growing popularity of3G(3rd-generation), Wi-Fi, LTE and other communication technologies. Thedevelopment of a variety of wireless technologies, in other words, thewireless heterogeneous networks, has promoted the terminal devices withmultiple interfaces. For example, many mobile devices like smart phones,laptops, and tablet PCs, are equipped with Cellular, Wi-Fi, and Bluetoothinterfaces. People gradually found themselves surrounded by various wirelessnetworks, which have different characteristics. Therefore, how to enable theterminal seamless roaming between different networks more efficiently, oreven integrate wireless networks from different ISP(Internet Service Provider)domain so as to improve user experience, has become a hot topic in academiaand industry.Handoff and bandwidth aggregation face several challenges whenconsidering user mobility in wireless environments. On the one hand, they arecaused by different wireless characteristics and the network infrastructure. Acommon scenario is that a user with a multi-interface device is under thecellular network of the mobile operator and Wi-Fi network of his office orhome at the same time. Generally speaking, due to the difference of the wireless technology, the mobile terminal connects to each network withdifferent interfaces. So the handoff process is visible to the applications, andthe applications need deal with the handoff operation. But to achieve seamlesshandover, it needs to be transparent to applications so as to keep the on-goingsessions. And since the cellular network and the Wi-Fi network belong todifferent network domains, it is hard to achieve the goal of bandwidthaggregation on data link layer, which is the general idea.Challenges are also brought by the uncertainty of the user mobility. Adecision good for the present may soon become poor when user moves toanother place. Meanwhile, performing the costly handoff or integration toofrequently would make the user experience worse. Thus, it’s necessary tostudy the user’s mobile model in order to predict the location of the nextperiod, and then make the best long-term decision.In this paper, we propose a network architecture which supports networklayer seamless handover and bandwidth aggregation, and the optimal handoffalgorithm based on Markov decision process. Secondly we analyze a largeamount of historical mobility traces, extract the user mobility model, andverify the feasibility of location prediction, which provides theoretical supportfor the prediction-based decision-making. Based on this, we build a systemmodel based on the Markov Decision Process (MDP), and derived amobility-aware handoff algorithm. Finally, we build a local test network toverify the feasibility of seamless handover and the efficiency of bandwidthaggregation. We also conduct trace-driven simulation experiments andcomparative study shows our prediction-based handoff algorithm significantlyoutperforms than the typical algorithms.