Mobile Computing Middleware

Cecilia Mascolo, Licia Capra, Wolfgang EmmerichNETWORKING Tutorials 2002 · DOI

TL;DR

This paper surveys the landscape of middleware for mobile computing, highlighting the unique challenges posed by mobile devices (resource constraints, intermittent connectivity) compared to traditional fixed systems. It proposes a framework for understanding these differences and categorizes existing and emerging middleware solutions, arguing for a shift towards lightweight, asynchronous, and context-aware approaches.

What Problem Does This Paper Address?

The explosive growth of mobile devices like smartphones and PDAs, coupled with advancements in wireless networking, has opened up new possibilities for ubiquitous computing. However, designing applications for these devices presents significant challenges. Unlike their stationary counterparts, mobile devices are characterized by:

  • Resource Scarcity: Limited battery power, processing speed, and memory.
  • Intermittent Connectivity: Frequent and unpredictable network disconnections due to movement or coverage gaps.
  • Dynamic Environments: Rapid changes in network bandwidth, location, and available services.

Traditional middleware, designed for stable, resource-rich, and stationary environments, often fails to meet these demands. This paper addresses the need for a new generation of middleware specifically tailored for the unique constraints and dynamics of mobile computing.

Methods

The authors propose a framework to characterize distributed systems and their middleware, focusing on three key concepts:

  1. Type of Device: Fixed vs. Mobile (and their associated resource capabilities).
  2. Type of Network Connection: Permanent vs. Intermittent.
  3. Type of Execution Context: Static vs. Dynamic.

This framework is used to analyze the limitations of traditional middleware (object-oriented, message-oriented, transaction-oriented) in mobile settings. The paper then surveys and categorizes various mobile middleware approaches based on:

  • Type of Computational Load: Heavyweight vs. Lightweight.
  • Type of Communication Paradigm: Synchronous vs. Asynchronous.
  • Type of Context Representation: Transparency vs. Awareness.

The survey covers middleware adapted from traditional systems and novel systems designed specifically for mobile environments, including reflective, context-aware, data-sharing, and tuple-space-based middleware.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Mobility is Fundamentally Different: The core assumption of constant connectivity and abundant resources in traditional distributed systems is a poor fit for mobile environments.
  • Asynchronous Communication is Key: Intermittent connectivity necessitates communication paradigms that don't require both parties to be online simultaneously.
  • Context Awareness is Crucial: Mobile applications and middleware must be aware of and adapt to dynamic environmental factors (location, bandwidth, device state) to provide effective service.
  • Lightweight Middleware is Essential: Resource constraints on mobile devices demand middleware with minimal computational and memory footprints.
  • Transparency is Not Always Optimal: While traditional middleware prioritizes hiding complexity, mobile environments may benefit from applications being aware of context to make better decisions.
  • Heterogeneity is the Norm: Future mobile networks will likely involve a diverse range of devices and connection technologies, requiring adaptable middleware.

Key Results

The paper doesn't present empirical results from a specific system but rather provides a comprehensive survey and classification of existing mobile middleware. It highlights that:

  • Traditional middleware (e.g., CORBA, JMS) can be adapted for mobile settings, but often with significant limitations due to their inherent heavyweight nature and synchronous communication.
  • Novel middleware approaches are emerging that explicitly address mobile challenges. These include:
  • Reflective Middleware: Allows dynamic adaptation and context awareness.
  • Context-Aware Middleware: Systems like Gaia and Odyssey leverage context information for adaptation.
  • Data-Sharing Middleware: Systems like Coda, Odyssey, Bayou, and Xmiddle focus on managing data consistency in disconnected or intermittently connected environments.
  • Tuple-Space Middleware: Lime, TSpaces, and L2imbo offer asynchronous, decoupled communication suitable for mobile scenarios.
  • No single existing middleware solution fully addresses all the requirements of mobile computing.

Limitations and Future Work

Limitations:

  • The paper acknowledges that a comprehensive survey of all existing and potential approaches is impossible.
  • Some mobile middleware solutions target only specific aspects of mobility (e.g., location awareness).
  • Traditional middleware adaptations often struggle with the fundamental constraints of mobile devices.

Future Work:

  • Developing middleware that is lightweight, asynchronous, and context-aware.
  • Creating customizable middleware that can adapt to diverse devices, operating systems, and network technologies.
  • Integrating application-dependent information for better adaptation and resource management.
  • Improving service discovery mechanisms in highly dynamic and ad-hoc mobile environments.
  • Addressing security concerns, which are amplified in mobile and dynamic settings.
  • Exploring the combination of functionalities from different middleware classes to create more robust solutions.

Summary updated

BibTeX
@inproceedings{mascolo2002mobile,
  author = {Cecilia Mascolo and Licia Capra and Wolfgang Emmerich},
  title = {{Mobile Computing Middleware}},
  booktitle = {NETWORKING Tutorials},
  year = {2002},
  doi = {10.1007/3-540-36162-6\_2},
}
Figure