Space Diversity-Based Grant-Free Random Access for Critical and Non-Critical IoT Services

Rahif Kassab, Osvaldo Simeone, Andrea Munari, Federico ClazzerICC 2020 · DOI

TL;DR

This paper proposes a grant-free random access scheme for Internet of Things (IoT) networks that simultaneously supports critical and non-critical services. By leveraging space diversity through multiple access points (APs) and considering both collision and superposition models for message reception, the authors demonstrate how to optimize throughput and reliability for these diverse service requirements.

What Problem Does This Paper Address?

Future IoT networks will host a vast array of devices with vastly different needs, ranging from highly reliable, low-latency critical services (like industrial control) to less demanding non-critical services (like environmental monitoring). Traditional grant-based access methods become inefficient due to the overhead of establishing connections for the massive number of sporadic IoT transmissions. Grant-free access, like slotted ALOHA, offers a more efficient alternative but faces challenges in managing interference and ensuring reliability, especially when supporting multiple service types with conflicting requirements. This paper tackles the problem of efficiently and reliably supporting both critical and non-critical IoT services within a single grant-free random access framework, considering the complexities of interference and channel erasures.

Methods

The core of the paper lies in analyzing a system model where IoT devices transmit packets using a grant-free slotted ALOHA protocol over an erasure channel to multiple Access Points (APs). These APs then act as relays, forwarding received messages to a central Base Station (BS) over a shared wireless backhaul channel, also modeled as an erasure channel. The key novelty lies in:

  • Space Diversity: Utilizing multiple APs to provide spatial diversity, which helps mitigate the impact of erasures and collisions.
  • Critical vs. Non-Critical Services: Differentiating between critical and non-critical messages, with critical messages transmitted at a higher power to improve their reliability, albeit at the cost of potentially interfering more with non-critical messages.
  • Two Receiver Models:
    • Collision Model: Assumes that multiple messages of the same type arriving at an AP or the BS cause a destructive collision, meaning only one message can be successfully decoded.
  • Superposition Model: Allows for the decoding of multiple instances of the same message if they arrive from different APs within a specific time synchronization window. This model leverages the benefits of multiple APs more effectively.
  • Resource Allocation Schemes: Comparing two approaches:
  • Non-orthogonal: Both critical and non-critical services share the same time slots.
  • Inter-service TDMA: A fraction of time slots is dedicated to critical services, and the rest to non-critical services.

The authors derive analytical expressions for throughput and reliability for both services under these models and schemes.

Datasets

Not discussed in the paper. The evaluation is based on simulated system parameters.

Key Insights and Ideas

  • Non-orthogonal Access for Critical Services: Non-orthogonal resource allocation generally outperforms inter-service TDMA for critical services in terms of both throughput and reliability. This is because critical services benefit from having access to a larger pool of resources and the spatial diversity provided by multiple APs.
  • TDMA for Non-Critical Services: For non-critical services, inter-service TDMA can provide better throughput and reliability. This is attributed to the elimination of interference from higher-power critical messages, which is a significant advantage for less stringent reliability requirements.
  • Superposition Model Advantage: The superposition model, which allows for decoding of superimposed signals from multiple APs, offers significant gains over the collision model, especially when leveraging space diversity. This highlights the importance of receiver capabilities and AP coordination (even if limited) in enhancing performance.
  • Interplay of Erasure Probabilities: The effectiveness of increasing the number of APs is highly dependent on the channel erasure probabilities of the radio access ($\epsilon_1$) and backhaul ($\epsilon_2$) links. When $\epsilon_1$ is high and $\epsilon_2$ is low, more APs can be beneficial by providing spatial diversity. However, if $\epsilon_2$ is high, increasing APs can lead to more backhaul collisions, diminishing returns.
  • Trade-off between Throughput and Reliability: As the frame duration (number of time slots, T) increases, reliability generally improves due to fewer collisions. However, throughput can be penalized if the increased T leads to unused time slots.

Key Results

  • Fig. 2 shows that non-orthogonal access can achieve a significant non-critical throughput without sacrificing critical throughput, while TDMA leads to a reduction in critical throughput for any increase in non-critical throughput.
  • Fig. 3 demonstrates that the optimal number of APs depends on the balance between radio access and backhaul channel conditions. For instance, with high radio access erasure probability ($\epsilon_1$) and low backhaul erasure probability ($\epsilon_2$), increasing APs initially helps but can eventually lead to backhaul collisions.
  • Fig. 4 illustrates that for critical services, non-orthogonal access outperforms TDMA in both throughput and reliability as the number of time slots (T) increases. Conversely, TDMA provides better throughput and reliability for non-critical services, as it avoids inter-service interference.

Limitations and Future Work

  • Collision Model Simplification: The paper acknowledges that the collision model is a simplification. Future work could consider a more general collision model where critical messages can tolerate a certain number of interfering non-critical messages.
  • Uncoordinated APs: The analysis assumes uncoordinated APs, which represents a worst-case scenario for dense deployments. Investigating the benefits of limited AP coordination could be a future direction.
  • Specific IoT Applications: While the paper categorizes services as critical and non-critical, future work could delve into the specific requirements of different IoT applications and tailor the access schemes accordingly.
  • Interference Management: Exploring more advanced interference management techniques beyond simple power control and spatial diversity could further enhance performance.

Summary updated

BibTeX
@inproceedings{kassab2020space,
  author = {Rahif Kassab and Osvaldo Simeone and Andrea Munari and Federico Clazzer},
  title = {{Space Diversity-Based Grant-Free Random Access for Critical and Non-Critical IoT Services}},
  booktitle = {ICC},
  year = {2020},
  doi = {10.1109/icc40277.2020.9148807},
}
Figure