Improving Information Freshness via Backbone-Assisted Cooperative Access Points

Haoyuan Pan, Yu Zhou, Tse-Tin Chan, Ming Tang, Jianqiang Li, Zhihua Du

This paper explores how to keep data "fresh" in dense Internet of Things (IoT) environments, such as smart cities, by allowing neighboring access points (APs) to cooperate over a wired backbone.

TL;DR

The researchers propose Soft-Co-AP, a system where neighboring access points share "soft information" (quantized probabilities) over an Ethernet backbone to jointly decode sensor data. This approach reduces the Age of Information (AoI) by up to 50% compared to single-AP systems and outperforms basic packet-sharing methods by finding the "Goldilocks" balance between decoding accuracy and backbone network delay.

What Problem Does This Paper Address?

In a smart city, thousands of sensors (measuring traffic, temperature, or pollution) send periodic updates to access points mounted on lamp posts. However, wireless channels are fickle; shadowing and fading often cause an AP to fail to decode a packet. In traditional systems, that data is simply lost, and the "Age of Information" (the time since the last successful update) grows until the next transmission.

While APs could cooperate to fix this, existing methods like Coordinated Multi-Point (CoMP) usually focus on maximizing total data throughput, not "freshness." Furthermore, sharing raw wireless signals over a standard Ethernet backbone creates massive traffic and high delays, which can actually make the information staler by the time it is finally decoded. The authors tackle the challenge of how to use backbone-assisted cooperation specifically to minimize AoI.

Methods

The authors introduce and compare three levels of system architecture:

  1. Single-AP: The baseline where each sensor communicates with only one primary AP.
  2. Co-AP (Cooperative AP): If the primary AP fails to decode a packet but a secondary neighbor succeeds, the neighbor forwards the fully decoded packet over the backbone.
  3. Soft-Co-AP: If neither AP can decode the packet individually, the secondary AP forwards "soft bits"—quantized Log-Likelihood Ratios (LLRs)—to the primary AP. The primary AP then performs Maximum Ratio Combining (MRC) to attempt a joint decode.

The Quantization Tradeoff: A core contribution is the study of how many bits ($m$) should represent each soft bit. More bits improve the chance of a successful joint decode but increase the size of the data sent over the backbone, leading to higher network delay.

Datasets

The study does not use a static pre-existing dataset. Instead, the authors built a real-world experimental testbed:

  • Hardware: Universal Software Radio Peripheral (USRP) devices acting as sensors and APs.
  • Software: GNU Radio for the physical layer (OFDM, BPSK, and Viterbi decoding).
  • Trace-Driven Simulation: They collected 50,000 real physical-layer decoding outcomes across various Signal-to-Noise Ratio (SNR) pairs to drive their AoI simulations.
  • Backbone Delay Data: They measured real-world Ethernet backbone delays using the Linux PING command under high-traffic conditions to model the latency of sharing soft bits.

Key Insights and Ideas

  • Freshness $\neq$ Throughput: In many communication papers, "better" means "more bits per second." Here, "better" means "the receiver has an update that was generated as recently as possible."
  • The 4-Bit Sweet Spot: The authors found that using 4 quantization bits per soft bit is often optimal. Using 1 bit (hard slicing) loses too much information for joint decoding, while 8 bits (standard precision) doubles the backbone delay without providing a proportional boost in decoding success.
  • Backbone Delay is a Freshness Killer: If the backbone delay is longer than the time until the sensor's next scheduled transmission, the cooperation is useless. The "old" recovered packet would arrive after a "newer" packet is already being sent.
  • Spatial Diversity via Backbone: By connecting APs via Ethernet, the system effectively creates a "virtual MIMO" array that is much more resilient to local interference and obstacles.

Key Results

  • Massive Gains over Single-AP: Basic Co-AP (sharing only decoded packets) reduces the average AoI by approximately 50% when there are 10 sensors in the network.
  • Soft-Co-AP Advantage: When the number of sensors is large (e.g., 30), Soft-Co-AP further reduces the average AoI by 12% compared to basic Co-AP.
  • Reliability Boost: At low SNR (5dB), cooperation increases the packet decoding probability by about 30%.
  • Scalability: Adding more than three cooperative APs provides diminishing returns, as the added backbone delay begins to outweigh the gains from increased spatial diversity.

Limitations and Future Work

  • TDMA Constraints: The current work assumes a Time Division Multiple Access (TDMA) schedule where sensors take turns. The authors acknowledge that real-world Wi-Fi uses random access (CSMA/CA), which introduces collision-related delays not fully explored here.
  • Two-AP Focus: While they briefly tested up to four APs, the mathematical modeling primarily focuses on a two-AP scenario.
  • Future Directions: The authors suggest extending this framework to Non-Orthogonal Multiple Access (NOMA) and more complex "cell-free" massive MIMO architectures where many APs serve many users simultaneously.

Summary updated

Abstract

Information freshness, characterized by Age of Information (AoI), is important for sensor applications involving timely status updates. In many cases, the wireless signals from one sensor can be received by multiple access points (APs). This article investigates the average AoI for cooperative APs (Co-APs), in which they can share information through a wired backbone network. We first study a basic backbone-assisted Co-AP system where APs share only decoded packets. Experimental results on software-defined radios (SDRs) indicate that Co-AP significantly improves the average AoI performance over a single-AP system. Next, we investigate an improved Co-AP system, called Soft-Co-AP. In addition to sharing decoded packets, Soft-Co-AP shares and collects soft information of packets that the APs fail to decode for further joint decoding. A critical issue in Soft-Co-AP is determining the number of quantization bits that represent the soft information (each soft bit) shared over the backbone. While more quantization bits per soft bit improves the joint decoding performance, it leads to higher backbone delay. We experimentally study the average AoI of Soft-Co-AP by evaluating the tradeoff between the backbone delay and the number of quantization bits. SDR experiments show that when the number of sensors is large, Soft-Co-AP further reduces the average AoI by 12% compared with Co-AP. Interestingly, good average AoI performance is usually achieved when the number of quantization bits per soft bit is neither too large nor too small.
BibTeX
@article{pan2023improving,
  author = {Haoyuan Pan and Yu Zhou and Tse-Tin Chan and Ming Tang and Jianqiang Li and Zhihua Du},
  title = {{Improving Information Freshness via Backbone-Assisted Cooperative Access Points}},
  journal = {IEEE Internet of Things Journal},
  year = {2023},
  doi = {10.1109/jiot.2022.3218948},
}

Loading related content...

Figure