Nodes use Waku's Proof Of Work algorithm to deter denial of service and various spam/flood attacks against the Whisper network. The sender of a message must perform some work which in this case means processing time. Because Status' main client is a mobile client, this easily leads to battery draining and poor performance of the app itself. Hence, all clients MUST use the following Whisper node settings:
| `topic-interest` | `0x05` | `[10000][4]byte` | Topic interest is used to share a node's interest in envelopes with specific topics. It does this in a more bandwidth considerate way, at the expense of some metadata protection. Peers MUST only send envelopes with specified topics. | [WAKU-1#topic-interest](https://github.com/vacp2p/specs/blob/master/specs/waku/waku-1.md#topic-interest-field), [the theoretical scaling model](https://github.com/vacp2p/research/tree/dcc71f4779be832d3b5ece9c4e11f1f7ec24aac2/whisper_scalability) |
<!-- TODO Add `light-node` and `confirmations-enabled` links when https://github.com/vacp2p/specs/pull/128 is merged -->
## Rate limiting
In order to provide an optional very basic Denial-of-Service attack protection, each node SHOULD define its own rate limits. The rate limits SHOULD be applied on IPs, peer IDs, and envelope topics.
Each node MAY decide to whitelist, i.e. do not rate limit, selected IPs or peer IDs.
If a peer exceeds node's rate limits, the connection between them MAY be dropped.
Each node SHOULD broadcast its rate limits to its peers using `rate limits` in `status-options` via packet code `0x00` or `0x22`. The rate limits is RLP-encoded information:
```
[ IP limits, PeerID limits, Topic limits ]
```
`IP limits`: 4-byte wide unsigned integer
`PeerID limits`: 4-byte wide unsigned integer
`Topic limits`: 4-byte wide unsigned integer
The rate limits MAY also be sent as an optional parameter in the handshake.
Each node SHOULD respect rate limits advertised by its peers. The number of packets SHOULD be throttled in order not to exceed peer's rate limits. If the limit gets exceeded, the connection MAY be dropped by the peer.
## Keys management
The protocol requires a key (symmetric or asymmetric) for the following actions:
* signing & verifying messages (asymmetric key)
* encrypting & decrypting messages (asymmetric or symmetric key).
Waku is broadcast-based protocol. In theory, everyone could communicate using a single topic but that would be extremely inefficient. Opposite would be using a unique topic for each conversation, however, this brings privacy concerns because it would be much easier to detect whether and when two parties have an active conversation.
Nodes use partitioned topics to broadcast private messages efficiently. By selecting a number of topic, it is possible to balance efficiency and privacy.
var hash []byte = keccak256(personalDiscoveryTopic)
var topicLen int = 4
if len(hash) <topicLen{
topicLen = len(hash)
}
var topic [4]byte
for i = 0; i <topicLen;i++{
topic[i] = hash[i]
}
```
Each Status Client SHOULD listen to this topic in order to receive ??? -->
### Group chat topic
Group chats does not have a dedicated topic. All group chat messages (including membership updates) are sent as one-to-one messages to multiple recipients.
### Negotiated topic
When a client sends a one to one message to another client, it MUST listen to their negotiated topic. This is computed by generating
a diffie-hellman key exchange between two members and taking the first four bytes of the `SHA3-256` of the key generated.
Even though, the protocol specifies an encryption layer that encrypts messages before passing them to the transport layer, Waku protocol requires each Waku message to be encrypted anyway.
The node encrypts public and group messages using symmetric encryption, and creates the key from a channel name string. The implementation is available in [`shh_generateSymKeyFromPassword`](https://github.com/ethereum/go-ethereum/wiki/Whisper-v6-RPC-API#shh_generatesymkeyfrompassword) JSON-RPC method of go-ethereum Whisper implementation.
Sending a message is a complex process where many things can go wrong. Message confirmations tell a node that a message originating from it has been seen by its direct peers.
A node MAY send a message confirmation for any batch of messages received in a packet Messages Code (`0x01`).
The Batch Acknowledge packet is followed by a keccak256 hash of the envelopes batch data (raw bytes).
The Message Response packet is more complex and is followed by a Versioned Message Response:
```
[ Version, Response]
```
`Version`: a version of the Message Response, equal to `1`,
`Response`: `[ Hash, Errors ]` where `Hash` is a keccak256 hash of the envelopes batch data (raw bytes) for which the confirmation is sent and `Errors` is a list of envelope errors when processing the batch. A single error contains `[ Hash, Code, Description ]` where `Hash` is a hash of the processed envelope, `Code` is an error code and `Description` is a descriptive error message.
The supported codes:
`1`: means time sync error which happens when an envelope is too old or created in the future (the root cause is no time sync between nodes).
The drawback of sending message confirmations is that it increases the noise in the network because for each sent message, one or more peers broadcast a corresponding confirmation. To limit that, both Batch Acknowledge packet (`0x0b`) and Message Response packet (`0x0c`) are not broadcast to peers of the peers, i.e. they do not follow epidemic spread.
In the current Status network setup, only `Mailservers` support message confirmations. A client posting a message to the network and after receiving a confirmation can be sure that the message got processed by the `Mailserver`. If additionally, sending a message is limited to non-`Mailserver` peers, it also guarantees that the message got broadcast through the network and it reached the selected `Mailserver`.
## Waku V1 extensions
### Request historic messages
Sends a request for historic messages to a `Mailserver`. The `Mailserver` node MUST be a direct peer and MUST be marked as trusted (using `waku_markTrustedPeer`).
The request does not wait for the response. It merely sends a peer-to-peer message to the `Mailserver` and it's up to `Mailserver` to process it and start sending historic messages.
The drawback of this approach is that it is impossible to tell which historic messages are the result of which request.
It's recommended to return messages from newest to oldest. To move further back in time, use `cursor` and `limit`.
