[Paper Review] Compressed Neighbor Discovery for Wireless Networks

Dongning Guo, Jun Luo+2 more|arXiv (Cornell University)|Dec 5, 2010

Wireless Networks and Protocols22 references19 citations

TL;DR

This paper proposes compressed neighbor discovery, a novel framework enabling all nodes in a wireless network to simultaneously identify their neighbors using a single transmission frame. By assigning unique on-off signatures and leveraging compressed sensing—via random group testing or deterministic Reed-Muller-based chirp decoding—nodes decode neighbors from signal superpositions, achieving 99.7% accuracy in a 1M-node network at 10 dB SNR with 4,096-symbol frames.

ABSTRACT

This paper studies neighbor discovery problem in wireless networks, namely, each node wishes to discover and identify the network interface addresses (NIAs) of those nodes within a single hop. A novel paradigm, called compressed neighbor discovery is proposed, which enables all nodes to simultaneously discover their respective neighborhoods with a single frame of transmission. The key technique is to assign each node a unique on-off signature and let all nodes simultaneously transmit their signatures. Despite that the radios are half-duplex, each node observes a superposition of its neighbors' signatures through its own off-slots. To identify its neighbors out of a large network address space, each node solves a compressed sensing problem. Two practical schemes are studied. The first employs random on-off signatures, and each node discovers its neighbors using a noncoherent detection algorithm based on group testing. The second scheme uses on-off signatures based on a deterministic second-order Reed-Muller code, and applies a chirp decoding algorithm. The second scheme needs much lower signal-to-noise ratio (SNR) and a shorter frame to achieve the same error performance. A network of over one million Poisson distributed nodes is studied numerically, where each node has 30 neighbors on average, and the channel between each pair of nodes is subject to path loss and Rayleigh fading. Within a single frame of 4,096 symbols, nodes can discover their respective neighbors with on average 99.7% accuracy at 10 dB SNR. The complexity of the chirp decoding algorithm is sub-linear, so that it is in principle scalable to networks with 48-bit IEEE 802.11 MAC addresses. The compressed neighbor discovery schemes are much more efficient than conventional random access discovery, where nodes have to retransmit over many frames with random delays to be successfully discovered due to collisions.

Motivation & Objective

• Address the inefficiency of conventional neighbor discovery in large-scale wireless networks, where random retransmissions lead to collisions and high latency.
• Overcome the limitations of half-duplex radios in detecting neighbors simultaneously without coordination.
• Enable scalable, low-complexity neighbor discovery in massive networks with large address spaces, such as those using 48-bit MAC addresses.
• Minimize the number of transmission frames and required SNR for reliable neighbor identification.
• Design practical schemes that are both efficient and scalable to real-world network sizes and channel conditions.

Proposed method

• Assign each node a unique on-off signature, where 'on' slots correspond to transmission and 'off' to listening.
• All nodes simultaneously transmit their signatures, and each node observes a superposition of neighbors' signals during its own off-slots.
• Model neighbor discovery as a compressed sensing problem, where the goal is to recover the set of active neighbors from noisy linear measurements.
• Implement two schemes: one using random on-off signatures with noncoherent group testing for neighbor detection.
• Implement a second scheme using deterministic second-order Reed-Muller codes for signatures, enabling chirp decoding with lower SNR requirements.
• Leverage the sub-linear complexity of the chirp decoding algorithm to ensure scalability to networks with 48-bit IEEE 802.11 MAC addresses.

Experimental results

Research questions

• RQ1 Can neighbor discovery be achieved in a single transmission frame without coordination, even with half-duplex radios?
• RQ2 How can compressed sensing be effectively applied to neighbor discovery in large-scale wireless networks with random node distributions?
• RQ3 What is the performance trade-off between random on-off signatures and deterministic codes in terms of SNR and frame length?
• RQ4 Can the proposed schemes scale to networks with 48-bit MAC addresses while maintaining low computational complexity?
• RQ5 How does the system perform under realistic channel conditions, including path loss and Rayleigh fading?

Key findings

• In a network of over one million Poisson-distributed nodes, each with an average of 30 neighbors, the system achieves 99.7% neighbor discovery accuracy within a single 4,096-symbol frame at 10 dB SNR.
• The deterministic Reed-Muller-based scheme outperforms the random signature scheme by requiring significantly lower SNR and shorter frame lengths for the same error performance.
• The chirp decoding algorithm achieves sub-linear computational complexity, making it feasible for scalability to networks with 48-bit IEEE 802.11 MAC addresses.
• The proposed compressed neighbor discovery framework reduces the number of required frames from many (in conventional random access) to just one, eliminating collision-related retransmissions.
• The system remains robust under realistic fading and path loss conditions, demonstrating practical viability in dense wireless environments.