1. 网络的可扩展性
2. 通信的可靠性
3. 高动态特性（不稳定拓扑）

# 1 未来网络特点分析

• Devices will be grouped by an inherent set of "interests" that are dependent on the tasks they are performing, and these group members will wish to communicate reliably between one another. Devices are not limited to a single group, and can belong to multiple groups.
• The majority of message exchanges will be within some local area, and long-distance traffic will only be a small fraction of overall communications.
• Any device can be a source or a sink, and traffic patterns between them may be one-to-one, one-to-many, many-to-one, or many-to-many.
• Future wireless environments will have a mix of mobile and stationary devices, where mobility will be typically be limited to some local area.

• No link state or neighbor information is utilized or maintained, and minimal control information is exchanged.
• Data is efficiently disseminated only across the region where group members exist. To support this, we develop a novel Group Discovery algorithm that dynamically discovers the region of interest and efficiently selects the minimal amount of relay nodes required to “cover” this region
• More reliable communications is achieved in an error-prone and mobile environment by using tunable resiliency, where the number of redundant data relays is configurable and is able to self-adjust in response to real-time channel conditions.
• Devices communicate in a many-to-many traffic pattern. Efficient one-to-one, one-to-many, and many-to-one are subsets.

# 2 GCN 机制内容

GCN 设计的目的是服务于设备分组聚集的网络，其中大部分的通信内容都在组内部进行。 网络中的节点资源受限，环境中存在干扰导致通信并不总是可靠。因通信需要：i) 能够抗干扰；ii) 高效实用带宽和能量。

## 2.1 业务模式

GCN 的一个网络布局的例子如下图：

GCN 的核心机制如下：

• Group Discovery: Efficient discovery of the local region where group members reside via a group discovery algorithm that is able to connect group members without the use of global control information
• Tunable resiliency: Relay nodes are activated such that the local region is sufficiently "covered" in data by having a tunable number of redundant data relays. This allows for resiliency towards both packet loss and mobility without the need for the constant exchange of control information. The number of activated relay nodes self-adjusts in response to real-time channel conditions
• Targeted flooding: Data can be efficiently and resiliently sent between sets of group members through an approach we call "targeted flooding"

## 2.2 Group Discovery

Group Discovery 的作用是在没有先验信息的情况下于局部发现彼此相连的组成员（不局限于一跳，但是最大跳跃数不会太高）。且这一过程不需要进行低效率的全局广播过程。Naive 的做法是使用 TTL 机制来做洪泛广播，不过这种方式非常低效，浪费通信资源。

1. 某个组成员通过发送一个 discovery message 来发起 group discovery 操作。相应的 discovery message 的 TTL 值，会根据 group 想要扩展的范围来进行设置。将最初发起者设置的 TTL 成为source TTL
2. 如果某个组成员监听到了一个 discovery message，该节点会使用 source TTL 来重新生成一个消息。
3. 如果一个非组成员监听到一个 discovery message，如果其 TTL 大于 0，那么将 TTL 减一后再次广播，反之如果 TTL 为 0，则什么都不做。

$N$个用户分布在面积为$A$的二维区域中，那么用户密度为$\lambda = N / A$。任意一个用户是组成员的概率是$P_g$，每个成员的通信传输距离都是$X$。在如上假设下，我们可以得到被发现的组成员的比例，同 TTL 初始值$T$的关系为$1 - e^{P_g \lambda \pi ((X - \frac{1}{2\lambda})\cdot T)^2}$。下图是仿真和理论结果的拟合情况。

## 2.3 Tunable Resiliency

1. 在 ACKACK 为组成员回复给潜在的中继节点的机发送前增加一小段延时。
2. 每个用户都记录其周围的可见邻居节点的数量，以方便确定其周围存在的可能的中继节点的数量。
3. 以随机性的方法选择一组用户用户依照一定的概率自行决策成为中继节点。

## 2.4 Targeted flooding

### 2.4.3 many-to-many traffic

many-to-many 过程中，各个发送节点独立发送 one-to-many 模式的消息即可。

# Reference

[1]
G. Kuperman, J. Sun, B.-N. Cheng, P. Deutsch, and A. Narula-Tam, “Group centric networking: A new approach for wireless multi-hop networking,” Ad Hoc Networks, vol. 79, pp. 160–172, 2018.
[2]
R. Ramanathan, R. Allan, P. Basu, J. Feinberg, G. Jakllari, V. Kawadia, S. Loos, J. Redi, C. Santivanez, and J. Freebersyser, “Scalability of mobile ad hoc networks: Theory vs practice,” in 2010-MILCOM 2010 MILITARY COMMUNICATIONS CONFERENCE, 2010, pp. 493–498.