Multipurpose mobility services for the Future Internet

Abstract

Mobility and handoff management is a key problem of the Future Internet. Current solutions provide mobility services, such as seamless mobility, adaptive mobility, and always-best-connected (ABC) mobility. The problem is these services achieve different purposes, work separately, and ignore conflicts between them; thus, they may improve one purpose and worsen others, yielding an erratic global behavior. In this article, we propose a new multipurpose mobility service that integrates multiple mobility services and supplies a fair balance between all the objectives to meet. As a proof-of-concept, we integrate the seamless, ABC, and adaptive mobility services, which have objectives in conflict. We formulate a multi-objective handoff optimization problem, which grades as NP-Hard. We develop a heuristic handoff algorithm, which provides near-optimal and balanced solutions. Finally, we evaluate the algorithm through random samples of simulated handoff scenarios, which provide hit rates over 90%.

Introduction

The area of Future Internet architectures is currently a hot and continuing research topic worldwide [1]. Many research projects such as Named Data Networking (NDN) [2], MobilityFirst [3], and Open Service Delivery Network (openSDN) [4] present architecture proposals intended to build the Future Internet. Each architecture proposal provides a different design of how the Internet could be in the future. We give our view of what features new architectures should have.

We envision the Future Internet as a communication system that is literally available anywhere, anytime (ubiquitous). The system connects a myriad of communicating entities or endpoints representing the original source or final destination of data, information, or contents. The communication system allows arbitrary pairs of endpoints to communicate (universal). The endpoints interact with a variety of terminals or end-devices. An initial end-device takes the information from the source entity (producer) and adapts it to enter the communication system. The communication system forwards the information to its final destination. A final end-device fetches the content from the system and adapts it to pass into the destination entity (consumer). The endpoints (users) and end-devices (terminals) are mobile entities, thus the communication system supports mobility across a variety of wireless/wired access networks (mobile). The system conveys any service over any access network over any terminal (multiservice). Services may include, load balancing, fault tolerance, multihoming, mobility management, strong security, energy-management, performance optimization, and so on. Service delivery over the Internet requires developing an open and secure service delivery network (SDN) architecture [4] where a variety of telecommunication carriers, network operators, and service providers offer SDN services that can be used by many application service providers (ASPs). Finally, the Future Internet exhibits high multidimensional heterogeneity. In particular, the new architectures need to manage at least five dimensions of heterogeneity: (1) diversity on service providers, network operators, telecommunication carriers; (2) diversity on applications and services; (3) diversity on access network technologies; (4) diversity on terminals or end-devices; and, (5) diversity on users and mobility of endpoints. The new architectures should hide such inevitable heterogeneity from humans (seamless), but paradoxically this necessarily requires a homogeneous layer of protocol software (uniform), typically known by the new architectures as the “narrow waist” of the Internet. In summary, the Future Internet [5] is ubiquitous, universal, mobile, multiservice, seamless, and uniform.

The problem of mobility management [6] is to preserve the leading of packets from source to target, while source, target, or both, change their network Points of Attachment (PoAs). This problem is easy to describe, yet complex to solve. The packets delivery service must adapt to connectivity changes, while it satisfies continuity, correctness, and timeliness constraints. To face this complexity, mobility management divides into two problems: location management, which determines the route to reach the target at any time, and handoff management, which preserves the communications while end-systems change their attachment points. Location managers deal with mobility protocols [7]. Handoff managers deal with handoff algorithms [8]. We focus on mobility and handoff management.

Many mobility services have been proposed in the literature in order to face the mobility management problem; e.g., seamless mobility [9], autonomic mobility [10], adaptive mobility [11], always-best-connected (ABC) mobility [12], secure mobility [13], energy-efficient mobility [14], timely-effective mobility [15], and others [16]. The problem is these services work separately, attain a different purpose, and ignore conflicts between them. Consequently, one service might improve one aspect of the communications but worsen others, yielding an erratic and unbalanced system. These mobility solutions may be seamless but not ABC, ABC but not adaptive, adaptive but not secure, secure but not power-efficient, etc. This means that single-purpose mobility services are not enough to face the challenges of the Future Internet.

When we integrate several single-purpose mobility services into a multipurpose mobility service, the integrated objectives may conflict with each other. In this case, a new task of the handoff control manager is to optimize and maintain a fair balance between all the objectives to meet. Since that task is not easy to achieve, and it affects the global behavior of handoff algorithms, we propose a paradigm shift from single-purpose to multipurpose mobility [17].

Despite the broad literature on mobility and handoff management, multipurpose mobility and multiobjective handoff optimization remain largely unexplored. To fill this gap, we integrate ABC mobility, seamless mobility, and adaptive mobility. The recent book of Dutta and Schulzrinne [18] seems to support the idea that more research on multiobjective handover optimization is necessary for the Future Internet.

The purpose of ABC mobility is to keep users always connected to the best available access network. Thus, to realize ABC mobility is required a mechanism to select the most suitable PoA and maximize the dwelling-time in the best available connection (DTiB).

The purpose of seamless mobility is to preserve service continuity. This requires reducing the communication disruptions during handoffs, which implies to minimize the handoff latency, the cumulative handoff latency (CHoL), the number of executed handoffs (nEHO), the signaling traffic overload, and the rates of lost packets and delayed packets.

The purpose of adaptive mobility is to keep the success of all handoffs in any mobility scenario. This requires a mechanism to determine when a handoff succeeds or fails, and how to estimate the rate of successful handoffs in a given mobility scenario. Adaptive mobility intends to maximize the number of successful handoffs (nSHO).

Particularly, seamless mobility, ABC mobility, and adaptive mobility are mutually in conflict, as will be shown in Section 2.4, so tradeoffs between conflicting objectives make multipurpose mobility a difficult problem. We are concerned with improving and balancing these services as much as possible.

In this paper, we formalize a Multi-Objective Handoff Optimization Problem (MOHOP) addressed to maximize DTiB, minimize nEHO, and maximize nSHO. As far as we know, there are no prior efforts providing a formalization and solution to this problem. Moreover, we classify this problem as NP-Hard. Using deterministic heuristics, we propose a handoff algorithm that provides near-optimal and balanced solutions in polynomial time. To verify this algorithm, we use a simple handoff simulator that creates samples of mobility (handoff) scenarios, displays the algorithm's behavior, and measures handoff performance parameters: DTiB, nEHO, and nSHO. A statistical analysis on hundreds of random samples estimates relative frequencies of acceptable solutions over 90%.