diff --git a/consensus/parlia/parlia.go b/consensus/parlia/parlia.go
index aed1052c1f..7522f555ac 100644
--- a/consensus/parlia/parlia.go
+++ b/consensus/parlia/parlia.go
@@ -67,7 +67,7 @@ const (
 
 	systemRewardPercent = 4 // it means 1/2^4 = 1/16 percentage of gas fee incoming will be distributed to system
 
-	collectAdditionalVotesRewardRatio = float64(1) // ratio of additional reward for collecting more votes than needed
+	collectAdditionalVotesRewardRatio = 100 // ratio of additional reward for collecting more votes than needed, the denominator is 100
 )
 
 var (
@@ -1027,7 +1027,7 @@ func (p *Parlia) distributeFinalityReward(chain consensus.ChainHeaderReader, sta
 		}
 		quorum := cmath.CeilDiv(len(snap.Validators)*2, 3)
 		if validVoteCount > quorum {
-			accumulatedWeights[head.Coinbase] += uint64(float64(validVoteCount-quorum) * collectAdditionalVotesRewardRatio)
+			accumulatedWeights[head.Coinbase] += uint64((validVoteCount - quorum) * collectAdditionalVotesRewardRatio / 100)
 		}
 	}
 
@@ -1788,7 +1788,7 @@ func encodeSigHeader(w io.Writer, header *types.Header, chainId *big.Int) {
 		header.GasLimit,
 		header.GasUsed,
 		header.Time,
-		header.Extra[:len(header.Extra)-65], // this will panic if extra is too short, should check before calling encodeSigHeader
+		header.Extra[:len(header.Extra)-extraSeal], // this will panic if extra is too short, should check before calling encodeSigHeader
 		header.MixDigest,
 		header.Nonce,
 	})
diff --git a/eth/handler.go b/eth/handler.go
index f3e05b2240..88afa3afef 100644
--- a/eth/handler.go
+++ b/eth/handler.go
@@ -944,6 +944,8 @@ func (h *handler) voteBroadcastLoop() {
 	for {
 		select {
 		case event := <-h.voteCh:
+			// The timeliness of votes is very important,
+			// so one vote will be sent instantly without waiting for other votes for batch sending by design.
 			h.BroadcastVote(event.Vote)
 		case <-h.votesSub.Err():
 			return
diff --git a/eth/handler_bsc.go b/eth/handler_bsc.go
index 195691989b..d01a475ed0 100644
--- a/eth/handler_bsc.go
+++ b/eth/handler_bsc.go
@@ -61,9 +61,15 @@ func (h *bscHandler) Handle(peer *bsc.Peer, packet bsc.Packet) error {
 // handleVotesBroadcast is invoked from a peer's message handler when it transmits a
 // votes broadcast for the local node to process.
 func (h *bscHandler) handleVotesBroadcast(peer *bsc.Peer, votes []*types.VoteEnvelope) error {
-	// Try to put votes into votepool
-	for _, vote := range votes {
-		h.votepool.PutVote(vote)
+	if peer.IsOverLimitAfterReceiving() {
+		peer.Log().Warn("peer sending votes too much, votes dropped; it may be a ddos attack, please check!")
+		return nil
 	}
+	// Here we only put the first vote, to avoid ddos attack by sending a large batch of votes.
+	// This won't abandon any valid vote, because one vote is sent every time referring to func voteBroadcastLoop
+	if len(votes) > 0 {
+		h.votepool.PutVote(votes[0])
+	}
+
 	return nil
 }
diff --git a/eth/protocols/bsc/peer.go b/eth/protocols/bsc/peer.go
index 202502a4b8..77ac11599f 100644
--- a/eth/protocols/bsc/peer.go
+++ b/eth/protocols/bsc/peer.go
@@ -1,6 +1,8 @@
 package bsc
 
 import (
+	"time"
+
 	mapset "github.com/deckarep/golang-set"
 
 	"github.com/ethereum/go-ethereum/common"
@@ -16,6 +18,15 @@ const (
 
 	// voteBufferSize is the maximum number of batch votes can be hold before sending
 	voteBufferSize = 21 * 2
+
+	// used to avoid of DDOS attack
+	// It's the max number of received votes per second from one peer
+	// 21 validators exist now, so 21 votes will be produced every one block interval
+	// so the limit is 7 = 21/3, here set it to 10 with a buffer.
+	receiveRateLimitPerSecond = 10
+
+	// the time span of one period
+	secondsPerPeriod = float64(10)
 )
 
 // max is a helper function which returns the larger of the two given integers.
@@ -31,6 +42,8 @@ type Peer struct {
 	id            string                     // Unique ID for the peer, cached
 	knownVotes    *knownCache                // Set of vote hashes known to be known by this peer
 	voteBroadcast chan []*types.VoteEnvelope // Channel used to queue votes propagation requests
+	periodBegin   time.Time                  // Begin time of the latest period for votes counting
+	periodCounter uint                       // Votes number in the latest period
 
 	*p2p.Peer                   // The embedded P2P package peer
 	rw        p2p.MsgReadWriter // Input/output streams for bsc
@@ -47,6 +60,8 @@ func NewPeer(version uint, p *p2p.Peer, rw p2p.MsgReadWriter) *Peer {
 		id:            id,
 		knownVotes:    newKnownCache(maxKnownVotes),
 		voteBroadcast: make(chan []*types.VoteEnvelope, voteBufferSize),
+		periodBegin:   time.Now(),
+		periodCounter: 0,
 		Peer:          p,
 		rw:            rw,
 		version:       version,
@@ -114,6 +129,18 @@ func (p *Peer) AsyncSendVotes(votes []*types.VoteEnvelope) {
 	}
 }
 
+// Step into the next period when secondsPerPeriod seconds passed,
+// Otherwise, check whether the number of received votes extra (secondsPerPeriod * receiveRateLimitPerSecond)
+func (p *Peer) IsOverLimitAfterReceiving() bool {
+	if timeInterval := time.Since(p.periodBegin).Seconds(); timeInterval >= secondsPerPeriod {
+		p.periodBegin = time.Now()
+		p.periodCounter = 0
+		return false
+	}
+	p.periodCounter += 1
+	return p.periodCounter > uint(secondsPerPeriod*receiveRateLimitPerSecond)
+}
+
 // broadcastVotes is a write loop that schedules votes broadcasts
 // to the remote peer. The goal is to have an async writer that does not lock up
 // node internals and at the same time rate limits queued data.